raid5.c 158.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4
/*
 * raid5.c : Multiple Devices driver for Linux
 *	   Copyright (C) 1996, 1997 Ingo Molnar, Miguel de Icaza, Gadi Oxman
 *	   Copyright (C) 1999, 2000 Ingo Molnar
5
 *	   Copyright (C) 2002, 2003 H. Peter Anvin
L
Linus Torvalds 已提交
6
 *
7 8 9
 * RAID-4/5/6 management functions.
 * Thanks to Penguin Computing for making the RAID-6 development possible
 * by donating a test server!
L
Linus Torvalds 已提交
10 11 12 13 14 15 16 17 18 19 20
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * You should have received a copy of the GNU General Public License
 * (for example /usr/src/linux/COPYING); if not, write to the Free
 * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

21 22 23 24 25 26 27 28 29
/*
 * BITMAP UNPLUGGING:
 *
 * The sequencing for updating the bitmap reliably is a little
 * subtle (and I got it wrong the first time) so it deserves some
 * explanation.
 *
 * We group bitmap updates into batches.  Each batch has a number.
 * We may write out several batches at once, but that isn't very important.
30 31
 * conf->seq_write is the number of the last batch successfully written.
 * conf->seq_flush is the number of the last batch that was closed to
32 33 34
 *    new additions.
 * When we discover that we will need to write to any block in a stripe
 * (in add_stripe_bio) we update the in-memory bitmap and record in sh->bm_seq
35
 * the number of the batch it will be in. This is seq_flush+1.
36 37 38 39 40 41 42 43 44
 * When we are ready to do a write, if that batch hasn't been written yet,
 *   we plug the array and queue the stripe for later.
 * When an unplug happens, we increment bm_flush, thus closing the current
 *   batch.
 * When we notice that bm_flush > bm_write, we write out all pending updates
 * to the bitmap, and advance bm_write to where bm_flush was.
 * This may occasionally write a bit out twice, but is sure never to
 * miss any bits.
 */
L
Linus Torvalds 已提交
45

46
#include <linux/blkdev.h>
47
#include <linux/kthread.h>
48
#include <linux/raid/pq.h>
49
#include <linux/async_tx.h>
50
#include <linux/async.h>
51
#include <linux/seq_file.h>
52
#include <linux/cpu.h>
53
#include <linux/slab.h>
54
#include <linux/ratelimit.h>
55
#include "md.h"
56
#include "raid5.h"
57
#include "raid0.h"
58
#include "bitmap.h"
59

L
Linus Torvalds 已提交
60 61 62 63 64 65 66 67 68
/*
 * Stripe cache
 */

#define NR_STRIPES		256
#define STRIPE_SIZE		PAGE_SIZE
#define STRIPE_SHIFT		(PAGE_SHIFT - 9)
#define STRIPE_SECTORS		(STRIPE_SIZE>>9)
#define	IO_THRESHOLD		1
69
#define BYPASS_THRESHOLD	1
70
#define NR_HASH			(PAGE_SIZE / sizeof(struct hlist_head))
L
Linus Torvalds 已提交
71 72
#define HASH_MASK		(NR_HASH - 1)

73 74 75 76 77
static inline struct hlist_head *stripe_hash(raid5_conf_t *conf, sector_t sect)
{
	int hash = (sect >> STRIPE_SHIFT) & HASH_MASK;
	return &conf->stripe_hashtbl[hash];
}
L
Linus Torvalds 已提交
78 79 80 81 82 83 84

/* bio's attached to a stripe+device for I/O are linked together in bi_sector
 * order without overlap.  There may be several bio's per stripe+device, and
 * a bio could span several devices.
 * When walking this list for a particular stripe+device, we must never proceed
 * beyond a bio that extends past this device, as the next bio might no longer
 * be valid.
85
 * This function is used to determine the 'next' bio in the list, given the sector
L
Linus Torvalds 已提交
86 87
 * of the current stripe+device
 */
88 89 90 91 92 93 94 95
static inline struct bio *r5_next_bio(struct bio *bio, sector_t sector)
{
	int sectors = bio->bi_size >> 9;
	if (bio->bi_sector + sectors < sector + STRIPE_SECTORS)
		return bio->bi_next;
	else
		return NULL;
}
L
Linus Torvalds 已提交
96

97
/*
98 99
 * We maintain a biased count of active stripes in the bottom 16 bits of
 * bi_phys_segments, and a count of processed stripes in the upper 16 bits
100 101 102
 */
static inline int raid5_bi_phys_segments(struct bio *bio)
{
103
	return bio->bi_phys_segments & 0xffff;
104 105 106 107
}

static inline int raid5_bi_hw_segments(struct bio *bio)
{
108
	return (bio->bi_phys_segments >> 16) & 0xffff;
109 110 111 112 113 114 115 116 117 118 119 120 121
}

static inline int raid5_dec_bi_phys_segments(struct bio *bio)
{
	--bio->bi_phys_segments;
	return raid5_bi_phys_segments(bio);
}

static inline int raid5_dec_bi_hw_segments(struct bio *bio)
{
	unsigned short val = raid5_bi_hw_segments(bio);

	--val;
122
	bio->bi_phys_segments = (val << 16) | raid5_bi_phys_segments(bio);
123 124 125 126 127
	return val;
}

static inline void raid5_set_bi_hw_segments(struct bio *bio, unsigned int cnt)
{
128
	bio->bi_phys_segments = raid5_bi_phys_segments(bio) | (cnt << 16);
129 130
}

131 132 133
/* Find first data disk in a raid6 stripe */
static inline int raid6_d0(struct stripe_head *sh)
{
134 135 136 137
	if (sh->ddf_layout)
		/* ddf always start from first device */
		return 0;
	/* md starts just after Q block */
138 139 140 141 142
	if (sh->qd_idx == sh->disks - 1)
		return 0;
	else
		return sh->qd_idx + 1;
}
143 144 145 146 147
static inline int raid6_next_disk(int disk, int raid_disks)
{
	disk++;
	return (disk < raid_disks) ? disk : 0;
}
148

149 150 151 152 153
/* When walking through the disks in a raid5, starting at raid6_d0,
 * We need to map each disk to a 'slot', where the data disks are slot
 * 0 .. raid_disks-3, the parity disk is raid_disks-2 and the Q disk
 * is raid_disks-1.  This help does that mapping.
 */
154 155
static int raid6_idx_to_slot(int idx, struct stripe_head *sh,
			     int *count, int syndrome_disks)
156
{
157
	int slot = *count;
158

159
	if (sh->ddf_layout)
160
		(*count)++;
161
	if (idx == sh->pd_idx)
162
		return syndrome_disks;
163
	if (idx == sh->qd_idx)
164
		return syndrome_disks + 1;
165
	if (!sh->ddf_layout)
166
		(*count)++;
167 168 169
	return slot;
}

170 171 172 173 174 175 176 177
static void return_io(struct bio *return_bi)
{
	struct bio *bi = return_bi;
	while (bi) {

		return_bi = bi->bi_next;
		bi->bi_next = NULL;
		bi->bi_size = 0;
178
		bio_endio(bi, 0);
179 180 181 182
		bi = return_bi;
	}
}

L
Linus Torvalds 已提交
183 184
static void print_raid5_conf (raid5_conf_t *conf);

185 186 187 188 189 190 191
static int stripe_operations_active(struct stripe_head *sh)
{
	return sh->check_state || sh->reconstruct_state ||
	       test_bit(STRIPE_BIOFILL_RUN, &sh->state) ||
	       test_bit(STRIPE_COMPUTE_RUN, &sh->state);
}

192
static void __release_stripe(raid5_conf_t *conf, struct stripe_head *sh)
L
Linus Torvalds 已提交
193 194
{
	if (atomic_dec_and_test(&sh->count)) {
195 196
		BUG_ON(!list_empty(&sh->lru));
		BUG_ON(atomic_read(&conf->active_stripes)==0);
L
Linus Torvalds 已提交
197
		if (test_bit(STRIPE_HANDLE, &sh->state)) {
N
NeilBrown 已提交
198
			if (test_bit(STRIPE_DELAYED, &sh->state))
L
Linus Torvalds 已提交
199
				list_add_tail(&sh->lru, &conf->delayed_list);
N
NeilBrown 已提交
200 201
			else if (test_bit(STRIPE_BIT_DELAY, &sh->state) &&
				   sh->bm_seq - conf->seq_write > 0)
202
				list_add_tail(&sh->lru, &conf->bitmap_list);
N
NeilBrown 已提交
203
			else {
204
				clear_bit(STRIPE_BIT_DELAY, &sh->state);
L
Linus Torvalds 已提交
205
				list_add_tail(&sh->lru, &conf->handle_list);
206
			}
L
Linus Torvalds 已提交
207 208
			md_wakeup_thread(conf->mddev->thread);
		} else {
209
			BUG_ON(stripe_operations_active(sh));
L
Linus Torvalds 已提交
210 211 212 213 214 215
			if (test_and_clear_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
				atomic_dec(&conf->preread_active_stripes);
				if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD)
					md_wakeup_thread(conf->mddev->thread);
			}
			atomic_dec(&conf->active_stripes);
216 217
			if (!test_bit(STRIPE_EXPANDING, &sh->state)) {
				list_add_tail(&sh->lru, &conf->inactive_list);
L
Linus Torvalds 已提交
218
				wake_up(&conf->wait_for_stripe);
219 220
				if (conf->retry_read_aligned)
					md_wakeup_thread(conf->mddev->thread);
221
			}
L
Linus Torvalds 已提交
222 223 224
		}
	}
}
225

L
Linus Torvalds 已提交
226 227 228 229
static void release_stripe(struct stripe_head *sh)
{
	raid5_conf_t *conf = sh->raid_conf;
	unsigned long flags;
230

L
Linus Torvalds 已提交
231 232 233 234 235
	spin_lock_irqsave(&conf->device_lock, flags);
	__release_stripe(conf, sh);
	spin_unlock_irqrestore(&conf->device_lock, flags);
}

236
static inline void remove_hash(struct stripe_head *sh)
L
Linus Torvalds 已提交
237
{
238 239
	pr_debug("remove_hash(), stripe %llu\n",
		(unsigned long long)sh->sector);
L
Linus Torvalds 已提交
240

241
	hlist_del_init(&sh->hash);
L
Linus Torvalds 已提交
242 243
}

244
static inline void insert_hash(raid5_conf_t *conf, struct stripe_head *sh)
L
Linus Torvalds 已提交
245
{
246
	struct hlist_head *hp = stripe_hash(conf, sh->sector);
L
Linus Torvalds 已提交
247

248 249
	pr_debug("insert_hash(), stripe %llu\n",
		(unsigned long long)sh->sector);
L
Linus Torvalds 已提交
250

251
	hlist_add_head(&sh->hash, hp);
L
Linus Torvalds 已提交
252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
}


/* find an idle stripe, make sure it is unhashed, and return it. */
static struct stripe_head *get_free_stripe(raid5_conf_t *conf)
{
	struct stripe_head *sh = NULL;
	struct list_head *first;

	if (list_empty(&conf->inactive_list))
		goto out;
	first = conf->inactive_list.next;
	sh = list_entry(first, struct stripe_head, lru);
	list_del_init(first);
	remove_hash(sh);
	atomic_inc(&conf->active_stripes);
out:
	return sh;
}

272
static void shrink_buffers(struct stripe_head *sh)
L
Linus Torvalds 已提交
273 274 275
{
	struct page *p;
	int i;
276
	int num = sh->raid_conf->pool_size;
L
Linus Torvalds 已提交
277

278
	for (i = 0; i < num ; i++) {
L
Linus Torvalds 已提交
279 280 281 282
		p = sh->dev[i].page;
		if (!p)
			continue;
		sh->dev[i].page = NULL;
283
		put_page(p);
L
Linus Torvalds 已提交
284 285 286
	}
}

287
static int grow_buffers(struct stripe_head *sh)
L
Linus Torvalds 已提交
288 289
{
	int i;
290
	int num = sh->raid_conf->pool_size;
L
Linus Torvalds 已提交
291

292
	for (i = 0; i < num; i++) {
L
Linus Torvalds 已提交
293 294 295 296 297 298 299 300 301 302
		struct page *page;

		if (!(page = alloc_page(GFP_KERNEL))) {
			return 1;
		}
		sh->dev[i].page = page;
	}
	return 0;
}

303
static void raid5_build_block(struct stripe_head *sh, int i, int previous);
304 305
static void stripe_set_idx(sector_t stripe, raid5_conf_t *conf, int previous,
			    struct stripe_head *sh);
L
Linus Torvalds 已提交
306

307
static void init_stripe(struct stripe_head *sh, sector_t sector, int previous)
L
Linus Torvalds 已提交
308 309
{
	raid5_conf_t *conf = sh->raid_conf;
310
	int i;
L
Linus Torvalds 已提交
311

312 313
	BUG_ON(atomic_read(&sh->count) != 0);
	BUG_ON(test_bit(STRIPE_HANDLE, &sh->state));
314
	BUG_ON(stripe_operations_active(sh));
315

316
	pr_debug("init_stripe called, stripe %llu\n",
L
Linus Torvalds 已提交
317 318 319
		(unsigned long long)sh->sector);

	remove_hash(sh);
320

321
	sh->generation = conf->generation - previous;
322
	sh->disks = previous ? conf->previous_raid_disks : conf->raid_disks;
L
Linus Torvalds 已提交
323
	sh->sector = sector;
324
	stripe_set_idx(sector, conf, previous, sh);
L
Linus Torvalds 已提交
325 326
	sh->state = 0;

327 328

	for (i = sh->disks; i--; ) {
L
Linus Torvalds 已提交
329 330
		struct r5dev *dev = &sh->dev[i];

331
		if (dev->toread || dev->read || dev->towrite || dev->written ||
L
Linus Torvalds 已提交
332
		    test_bit(R5_LOCKED, &dev->flags)) {
333
			printk(KERN_ERR "sector=%llx i=%d %p %p %p %p %d\n",
L
Linus Torvalds 已提交
334
			       (unsigned long long)sh->sector, i, dev->toread,
335
			       dev->read, dev->towrite, dev->written,
L
Linus Torvalds 已提交
336
			       test_bit(R5_LOCKED, &dev->flags));
337
			WARN_ON(1);
L
Linus Torvalds 已提交
338 339
		}
		dev->flags = 0;
340
		raid5_build_block(sh, i, previous);
L
Linus Torvalds 已提交
341 342 343 344
	}
	insert_hash(conf, sh);
}

345 346
static struct stripe_head *__find_stripe(raid5_conf_t *conf, sector_t sector,
					 short generation)
L
Linus Torvalds 已提交
347 348
{
	struct stripe_head *sh;
349
	struct hlist_node *hn;
L
Linus Torvalds 已提交
350

351
	pr_debug("__find_stripe, sector %llu\n", (unsigned long long)sector);
352
	hlist_for_each_entry(sh, hn, stripe_hash(conf, sector), hash)
353
		if (sh->sector == sector && sh->generation == generation)
L
Linus Torvalds 已提交
354
			return sh;
355
	pr_debug("__stripe %llu not in cache\n", (unsigned long long)sector);
L
Linus Torvalds 已提交
356 357 358
	return NULL;
}

359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
/*
 * Need to check if array has failed when deciding whether to:
 *  - start an array
 *  - remove non-faulty devices
 *  - add a spare
 *  - allow a reshape
 * This determination is simple when no reshape is happening.
 * However if there is a reshape, we need to carefully check
 * both the before and after sections.
 * This is because some failed devices may only affect one
 * of the two sections, and some non-in_sync devices may
 * be insync in the section most affected by failed devices.
 */
static int has_failed(raid5_conf_t *conf)
{
	int degraded;
	int i;
	if (conf->mddev->reshape_position == MaxSector)
		return conf->mddev->degraded > conf->max_degraded;

	rcu_read_lock();
	degraded = 0;
	for (i = 0; i < conf->previous_raid_disks; i++) {
		mdk_rdev_t *rdev = rcu_dereference(conf->disks[i].rdev);
		if (!rdev || test_bit(Faulty, &rdev->flags))
			degraded++;
		else if (test_bit(In_sync, &rdev->flags))
			;
		else
			/* not in-sync or faulty.
			 * If the reshape increases the number of devices,
			 * this is being recovered by the reshape, so
			 * this 'previous' section is not in_sync.
			 * If the number of devices is being reduced however,
			 * the device can only be part of the array if
			 * we are reverting a reshape, so this section will
			 * be in-sync.
			 */
			if (conf->raid_disks >= conf->previous_raid_disks)
				degraded++;
	}
	rcu_read_unlock();
	if (degraded > conf->max_degraded)
		return 1;
	rcu_read_lock();
	degraded = 0;
	for (i = 0; i < conf->raid_disks; i++) {
		mdk_rdev_t *rdev = rcu_dereference(conf->disks[i].rdev);
		if (!rdev || test_bit(Faulty, &rdev->flags))
			degraded++;
		else if (test_bit(In_sync, &rdev->flags))
			;
		else
			/* not in-sync or faulty.
			 * If reshape increases the number of devices, this
			 * section has already been recovered, else it
			 * almost certainly hasn't.
			 */
			if (conf->raid_disks <= conf->previous_raid_disks)
				degraded++;
	}
	rcu_read_unlock();
	if (degraded > conf->max_degraded)
		return 1;
	return 0;
}

426 427
static struct stripe_head *
get_active_stripe(raid5_conf_t *conf, sector_t sector,
428
		  int previous, int noblock, int noquiesce)
L
Linus Torvalds 已提交
429 430 431
{
	struct stripe_head *sh;

432
	pr_debug("get_stripe, sector %llu\n", (unsigned long long)sector);
L
Linus Torvalds 已提交
433 434 435 436

	spin_lock_irq(&conf->device_lock);

	do {
437
		wait_event_lock_irq(conf->wait_for_stripe,
438
				    conf->quiesce == 0 || noquiesce,
439
				    conf->device_lock, /* nothing */);
440
		sh = __find_stripe(conf, sector, conf->generation - previous);
L
Linus Torvalds 已提交
441 442 443 444 445 446 447 448 449
		if (!sh) {
			if (!conf->inactive_blocked)
				sh = get_free_stripe(conf);
			if (noblock && sh == NULL)
				break;
			if (!sh) {
				conf->inactive_blocked = 1;
				wait_event_lock_irq(conf->wait_for_stripe,
						    !list_empty(&conf->inactive_list) &&
450 451
						    (atomic_read(&conf->active_stripes)
						     < (conf->max_nr_stripes *3/4)
L
Linus Torvalds 已提交
452 453
						     || !conf->inactive_blocked),
						    conf->device_lock,
454
						    );
L
Linus Torvalds 已提交
455 456
				conf->inactive_blocked = 0;
			} else
457
				init_stripe(sh, sector, previous);
L
Linus Torvalds 已提交
458 459
		} else {
			if (atomic_read(&sh->count)) {
460 461
				BUG_ON(!list_empty(&sh->lru)
				    && !test_bit(STRIPE_EXPANDING, &sh->state));
L
Linus Torvalds 已提交
462 463 464
			} else {
				if (!test_bit(STRIPE_HANDLE, &sh->state))
					atomic_inc(&conf->active_stripes);
465 466
				if (list_empty(&sh->lru) &&
				    !test_bit(STRIPE_EXPANDING, &sh->state))
467 468
					BUG();
				list_del_init(&sh->lru);
L
Linus Torvalds 已提交
469 470 471 472 473 474 475 476 477 478 479
			}
		}
	} while (sh == NULL);

	if (sh)
		atomic_inc(&sh->count);

	spin_unlock_irq(&conf->device_lock);
	return sh;
}

480 481 482 483
static void
raid5_end_read_request(struct bio *bi, int error);
static void
raid5_end_write_request(struct bio *bi, int error);
484

485
static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
486 487 488 489 490 491 492 493 494 495
{
	raid5_conf_t *conf = sh->raid_conf;
	int i, disks = sh->disks;

	might_sleep();

	for (i = disks; i--; ) {
		int rw;
		struct bio *bi;
		mdk_rdev_t *rdev;
T
Tejun Heo 已提交
496 497 498 499 500 501
		if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags)) {
			if (test_and_clear_bit(R5_WantFUA, &sh->dev[i].flags))
				rw = WRITE_FUA;
			else
				rw = WRITE;
		} else if (test_and_clear_bit(R5_Wantread, &sh->dev[i].flags))
502 503 504 505 506 507 508
			rw = READ;
		else
			continue;

		bi = &sh->dev[i].req;

		bi->bi_rw = rw;
509
		if (rw & WRITE)
510 511 512 513 514 515 516 517 518 519 520 521
			bi->bi_end_io = raid5_end_write_request;
		else
			bi->bi_end_io = raid5_end_read_request;

		rcu_read_lock();
		rdev = rcu_dereference(conf->disks[i].rdev);
		if (rdev && test_bit(Faulty, &rdev->flags))
			rdev = NULL;
		if (rdev)
			atomic_inc(&rdev->nr_pending);
		rcu_read_unlock();

522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551
		/* We have already checked bad blocks for reads.  Now
		 * need to check for writes.
		 */
		while ((rw & WRITE) && rdev &&
		       test_bit(WriteErrorSeen, &rdev->flags)) {
			sector_t first_bad;
			int bad_sectors;
			int bad = is_badblock(rdev, sh->sector, STRIPE_SECTORS,
					      &first_bad, &bad_sectors);
			if (!bad)
				break;

			if (bad < 0) {
				set_bit(BlockedBadBlocks, &rdev->flags);
				if (!conf->mddev->external &&
				    conf->mddev->flags) {
					/* It is very unlikely, but we might
					 * still need to write out the
					 * bad block log - better give it
					 * a chance*/
					md_check_recovery(conf->mddev);
				}
				md_wait_for_blocked_rdev(rdev, conf->mddev);
			} else {
				/* Acknowledged bad block - skip the write */
				rdev_dec_pending(rdev, conf->mddev);
				rdev = NULL;
			}
		}

552
		if (rdev) {
553
			if (s->syncing || s->expanding || s->expanded)
554 555
				md_sync_acct(rdev->bdev, STRIPE_SECTORS);

D
Dan Williams 已提交
556 557
			set_bit(STRIPE_IO_STARTED, &sh->state);

558 559
			bi->bi_bdev = rdev->bdev;
			pr_debug("%s: for %llu schedule op %ld on disc %d\n",
560
				__func__, (unsigned long long)sh->sector,
561 562 563 564 565 566 567 568 569 570 571 572 573 574
				bi->bi_rw, i);
			atomic_inc(&sh->count);
			bi->bi_sector = sh->sector + rdev->data_offset;
			bi->bi_flags = 1 << BIO_UPTODATE;
			bi->bi_vcnt = 1;
			bi->bi_max_vecs = 1;
			bi->bi_idx = 0;
			bi->bi_io_vec = &sh->dev[i].vec;
			bi->bi_io_vec[0].bv_len = STRIPE_SIZE;
			bi->bi_io_vec[0].bv_offset = 0;
			bi->bi_size = STRIPE_SIZE;
			bi->bi_next = NULL;
			generic_make_request(bi);
		} else {
575
			if (rw & WRITE)
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
				set_bit(STRIPE_DEGRADED, &sh->state);
			pr_debug("skip op %ld on disc %d for sector %llu\n",
				bi->bi_rw, i, (unsigned long long)sh->sector);
			clear_bit(R5_LOCKED, &sh->dev[i].flags);
			set_bit(STRIPE_HANDLE, &sh->state);
		}
	}
}

static struct dma_async_tx_descriptor *
async_copy_data(int frombio, struct bio *bio, struct page *page,
	sector_t sector, struct dma_async_tx_descriptor *tx)
{
	struct bio_vec *bvl;
	struct page *bio_page;
	int i;
	int page_offset;
593
	struct async_submit_ctl submit;
D
Dan Williams 已提交
594
	enum async_tx_flags flags = 0;
595 596 597 598 599

	if (bio->bi_sector >= sector)
		page_offset = (signed)(bio->bi_sector - sector) * 512;
	else
		page_offset = (signed)(sector - bio->bi_sector) * -512;
600

D
Dan Williams 已提交
601 602 603 604
	if (frombio)
		flags |= ASYNC_TX_FENCE;
	init_async_submit(&submit, flags, tx, NULL, NULL, NULL);

605
	bio_for_each_segment(bvl, bio, i) {
606
		int len = bvl->bv_len;
607 608 609 610 611 612 613 614 615 616 617 618 619 620 621
		int clen;
		int b_offset = 0;

		if (page_offset < 0) {
			b_offset = -page_offset;
			page_offset += b_offset;
			len -= b_offset;
		}

		if (len > 0 && page_offset + len > STRIPE_SIZE)
			clen = STRIPE_SIZE - page_offset;
		else
			clen = len;

		if (clen > 0) {
622 623
			b_offset += bvl->bv_offset;
			bio_page = bvl->bv_page;
624 625
			if (frombio)
				tx = async_memcpy(page, bio_page, page_offset,
626
						  b_offset, clen, &submit);
627 628
			else
				tx = async_memcpy(bio_page, page, b_offset,
629
						  page_offset, clen, &submit);
630
		}
631 632 633
		/* chain the operations */
		submit.depend_tx = tx;

634 635 636 637 638 639 640 641 642 643 644 645 646
		if (clen < len) /* hit end of page */
			break;
		page_offset +=  len;
	}

	return tx;
}

static void ops_complete_biofill(void *stripe_head_ref)
{
	struct stripe_head *sh = stripe_head_ref;
	struct bio *return_bi = NULL;
	raid5_conf_t *conf = sh->raid_conf;
647
	int i;
648

649
	pr_debug("%s: stripe %llu\n", __func__,
650 651 652
		(unsigned long long)sh->sector);

	/* clear completed biofills */
653
	spin_lock_irq(&conf->device_lock);
654 655 656 657
	for (i = sh->disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];

		/* acknowledge completion of a biofill operation */
658 659
		/* and check if we need to reply to a read request,
		 * new R5_Wantfill requests are held off until
660
		 * !STRIPE_BIOFILL_RUN
661 662
		 */
		if (test_and_clear_bit(R5_Wantfill, &dev->flags)) {
663 664 665 666 667 668 669 670
			struct bio *rbi, *rbi2;

			BUG_ON(!dev->read);
			rbi = dev->read;
			dev->read = NULL;
			while (rbi && rbi->bi_sector <
				dev->sector + STRIPE_SECTORS) {
				rbi2 = r5_next_bio(rbi, dev->sector);
671
				if (!raid5_dec_bi_phys_segments(rbi)) {
672 673 674 675 676 677 678
					rbi->bi_next = return_bi;
					return_bi = rbi;
				}
				rbi = rbi2;
			}
		}
	}
679 680
	spin_unlock_irq(&conf->device_lock);
	clear_bit(STRIPE_BIOFILL_RUN, &sh->state);
681 682 683

	return_io(return_bi);

684
	set_bit(STRIPE_HANDLE, &sh->state);
685 686 687 688 689 690 691
	release_stripe(sh);
}

static void ops_run_biofill(struct stripe_head *sh)
{
	struct dma_async_tx_descriptor *tx = NULL;
	raid5_conf_t *conf = sh->raid_conf;
692
	struct async_submit_ctl submit;
693 694
	int i;

695
	pr_debug("%s: stripe %llu\n", __func__,
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715
		(unsigned long long)sh->sector);

	for (i = sh->disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
		if (test_bit(R5_Wantfill, &dev->flags)) {
			struct bio *rbi;
			spin_lock_irq(&conf->device_lock);
			dev->read = rbi = dev->toread;
			dev->toread = NULL;
			spin_unlock_irq(&conf->device_lock);
			while (rbi && rbi->bi_sector <
				dev->sector + STRIPE_SECTORS) {
				tx = async_copy_data(0, rbi, dev->page,
					dev->sector, tx);
				rbi = r5_next_bio(rbi, dev->sector);
			}
		}
	}

	atomic_inc(&sh->count);
716 717
	init_async_submit(&submit, ASYNC_TX_ACK, tx, ops_complete_biofill, sh, NULL);
	async_trigger_callback(&submit);
718 719
}

720
static void mark_target_uptodate(struct stripe_head *sh, int target)
721
{
722
	struct r5dev *tgt;
723

724 725
	if (target < 0)
		return;
726

727
	tgt = &sh->dev[target];
728 729 730
	set_bit(R5_UPTODATE, &tgt->flags);
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
	clear_bit(R5_Wantcompute, &tgt->flags);
731 732
}

733
static void ops_complete_compute(void *stripe_head_ref)
734 735 736
{
	struct stripe_head *sh = stripe_head_ref;

737
	pr_debug("%s: stripe %llu\n", __func__,
738 739
		(unsigned long long)sh->sector);

740
	/* mark the computed target(s) as uptodate */
741
	mark_target_uptodate(sh, sh->ops.target);
742
	mark_target_uptodate(sh, sh->ops.target2);
743

744 745 746
	clear_bit(STRIPE_COMPUTE_RUN, &sh->state);
	if (sh->check_state == check_state_compute_run)
		sh->check_state = check_state_compute_result;
747 748 749 750
	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

751 752 753 754 755 756 757 758 759
/* return a pointer to the address conversion region of the scribble buffer */
static addr_conv_t *to_addr_conv(struct stripe_head *sh,
				 struct raid5_percpu *percpu)
{
	return percpu->scribble + sizeof(struct page *) * (sh->disks + 2);
}

static struct dma_async_tx_descriptor *
ops_run_compute5(struct stripe_head *sh, struct raid5_percpu *percpu)
760 761
{
	int disks = sh->disks;
762
	struct page **xor_srcs = percpu->scribble;
763 764 765 766 767
	int target = sh->ops.target;
	struct r5dev *tgt = &sh->dev[target];
	struct page *xor_dest = tgt->page;
	int count = 0;
	struct dma_async_tx_descriptor *tx;
768
	struct async_submit_ctl submit;
769 770 771
	int i;

	pr_debug("%s: stripe %llu block: %d\n",
772
		__func__, (unsigned long long)sh->sector, target);
773 774 775 776 777 778 779 780
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));

	for (i = disks; i--; )
		if (i != target)
			xor_srcs[count++] = sh->dev[i].page;

	atomic_inc(&sh->count);

D
Dan Williams 已提交
781
	init_async_submit(&submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_ZERO_DST, NULL,
782
			  ops_complete_compute, sh, to_addr_conv(sh, percpu));
783
	if (unlikely(count == 1))
784
		tx = async_memcpy(xor_dest, xor_srcs[0], 0, 0, STRIPE_SIZE, &submit);
785
	else
786
		tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE, &submit);
787 788 789 790

	return tx;
}

791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808
/* set_syndrome_sources - populate source buffers for gen_syndrome
 * @srcs - (struct page *) array of size sh->disks
 * @sh - stripe_head to parse
 *
 * Populates srcs in proper layout order for the stripe and returns the
 * 'count' of sources to be used in a call to async_gen_syndrome.  The P
 * destination buffer is recorded in srcs[count] and the Q destination
 * is recorded in srcs[count+1]].
 */
static int set_syndrome_sources(struct page **srcs, struct stripe_head *sh)
{
	int disks = sh->disks;
	int syndrome_disks = sh->ddf_layout ? disks : (disks - 2);
	int d0_idx = raid6_d0(sh);
	int count;
	int i;

	for (i = 0; i < disks; i++)
809
		srcs[i] = NULL;
810 811 812 813 814 815 816 817 818 819

	count = 0;
	i = d0_idx;
	do {
		int slot = raid6_idx_to_slot(i, sh, &count, syndrome_disks);

		srcs[slot] = sh->dev[i].page;
		i = raid6_next_disk(i, disks);
	} while (i != d0_idx);

820
	return syndrome_disks;
821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840
}

static struct dma_async_tx_descriptor *
ops_run_compute6_1(struct stripe_head *sh, struct raid5_percpu *percpu)
{
	int disks = sh->disks;
	struct page **blocks = percpu->scribble;
	int target;
	int qd_idx = sh->qd_idx;
	struct dma_async_tx_descriptor *tx;
	struct async_submit_ctl submit;
	struct r5dev *tgt;
	struct page *dest;
	int i;
	int count;

	if (sh->ops.target < 0)
		target = sh->ops.target2;
	else if (sh->ops.target2 < 0)
		target = sh->ops.target;
841
	else
842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
		/* we should only have one valid target */
		BUG();
	BUG_ON(target < 0);
	pr_debug("%s: stripe %llu block: %d\n",
		__func__, (unsigned long long)sh->sector, target);

	tgt = &sh->dev[target];
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
	dest = tgt->page;

	atomic_inc(&sh->count);

	if (target == qd_idx) {
		count = set_syndrome_sources(blocks, sh);
		blocks[count] = NULL; /* regenerating p is not necessary */
		BUG_ON(blocks[count+1] != dest); /* q should already be set */
D
Dan Williams 已提交
858 859
		init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
				  ops_complete_compute, sh,
860 861 862 863 864 865 866 867 868 869 870
				  to_addr_conv(sh, percpu));
		tx = async_gen_syndrome(blocks, 0, count+2, STRIPE_SIZE, &submit);
	} else {
		/* Compute any data- or p-drive using XOR */
		count = 0;
		for (i = disks; i-- ; ) {
			if (i == target || i == qd_idx)
				continue;
			blocks[count++] = sh->dev[i].page;
		}

D
Dan Williams 已提交
871 872
		init_async_submit(&submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_ZERO_DST,
				  NULL, ops_complete_compute, sh,
873 874 875
				  to_addr_conv(sh, percpu));
		tx = async_xor(dest, blocks, 0, count, STRIPE_SIZE, &submit);
	}
876 877 878 879

	return tx;
}

880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900
static struct dma_async_tx_descriptor *
ops_run_compute6_2(struct stripe_head *sh, struct raid5_percpu *percpu)
{
	int i, count, disks = sh->disks;
	int syndrome_disks = sh->ddf_layout ? disks : disks-2;
	int d0_idx = raid6_d0(sh);
	int faila = -1, failb = -1;
	int target = sh->ops.target;
	int target2 = sh->ops.target2;
	struct r5dev *tgt = &sh->dev[target];
	struct r5dev *tgt2 = &sh->dev[target2];
	struct dma_async_tx_descriptor *tx;
	struct page **blocks = percpu->scribble;
	struct async_submit_ctl submit;

	pr_debug("%s: stripe %llu block1: %d block2: %d\n",
		 __func__, (unsigned long long)sh->sector, target, target2);
	BUG_ON(target < 0 || target2 < 0);
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
	BUG_ON(!test_bit(R5_Wantcompute, &tgt2->flags));

901
	/* we need to open-code set_syndrome_sources to handle the
902 903 904
	 * slot number conversion for 'faila' and 'failb'
	 */
	for (i = 0; i < disks ; i++)
905
		blocks[i] = NULL;
906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931
	count = 0;
	i = d0_idx;
	do {
		int slot = raid6_idx_to_slot(i, sh, &count, syndrome_disks);

		blocks[slot] = sh->dev[i].page;

		if (i == target)
			faila = slot;
		if (i == target2)
			failb = slot;
		i = raid6_next_disk(i, disks);
	} while (i != d0_idx);

	BUG_ON(faila == failb);
	if (failb < faila)
		swap(faila, failb);
	pr_debug("%s: stripe: %llu faila: %d failb: %d\n",
		 __func__, (unsigned long long)sh->sector, faila, failb);

	atomic_inc(&sh->count);

	if (failb == syndrome_disks+1) {
		/* Q disk is one of the missing disks */
		if (faila == syndrome_disks) {
			/* Missing P+Q, just recompute */
D
Dan Williams 已提交
932 933 934
			init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
					  ops_complete_compute, sh,
					  to_addr_conv(sh, percpu));
935
			return async_gen_syndrome(blocks, 0, syndrome_disks+2,
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954
						  STRIPE_SIZE, &submit);
		} else {
			struct page *dest;
			int data_target;
			int qd_idx = sh->qd_idx;

			/* Missing D+Q: recompute D from P, then recompute Q */
			if (target == qd_idx)
				data_target = target2;
			else
				data_target = target;

			count = 0;
			for (i = disks; i-- ; ) {
				if (i == data_target || i == qd_idx)
					continue;
				blocks[count++] = sh->dev[i].page;
			}
			dest = sh->dev[data_target].page;
D
Dan Williams 已提交
955 956 957 958
			init_async_submit(&submit,
					  ASYNC_TX_FENCE|ASYNC_TX_XOR_ZERO_DST,
					  NULL, NULL, NULL,
					  to_addr_conv(sh, percpu));
959 960 961 962
			tx = async_xor(dest, blocks, 0, count, STRIPE_SIZE,
				       &submit);

			count = set_syndrome_sources(blocks, sh);
D
Dan Williams 已提交
963 964 965
			init_async_submit(&submit, ASYNC_TX_FENCE, tx,
					  ops_complete_compute, sh,
					  to_addr_conv(sh, percpu));
966 967 968 969
			return async_gen_syndrome(blocks, 0, count+2,
						  STRIPE_SIZE, &submit);
		}
	} else {
970 971 972 973 974 975 976 977 978 979 980 981 982 983
		init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
				  ops_complete_compute, sh,
				  to_addr_conv(sh, percpu));
		if (failb == syndrome_disks) {
			/* We're missing D+P. */
			return async_raid6_datap_recov(syndrome_disks+2,
						       STRIPE_SIZE, faila,
						       blocks, &submit);
		} else {
			/* We're missing D+D. */
			return async_raid6_2data_recov(syndrome_disks+2,
						       STRIPE_SIZE, faila, failb,
						       blocks, &submit);
		}
984 985 986 987
	}
}


988 989 990 991
static void ops_complete_prexor(void *stripe_head_ref)
{
	struct stripe_head *sh = stripe_head_ref;

992
	pr_debug("%s: stripe %llu\n", __func__,
993 994 995 996
		(unsigned long long)sh->sector);
}

static struct dma_async_tx_descriptor *
997 998
ops_run_prexor(struct stripe_head *sh, struct raid5_percpu *percpu,
	       struct dma_async_tx_descriptor *tx)
999 1000
{
	int disks = sh->disks;
1001
	struct page **xor_srcs = percpu->scribble;
1002
	int count = 0, pd_idx = sh->pd_idx, i;
1003
	struct async_submit_ctl submit;
1004 1005 1006 1007

	/* existing parity data subtracted */
	struct page *xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;

1008
	pr_debug("%s: stripe %llu\n", __func__,
1009 1010 1011 1012 1013
		(unsigned long long)sh->sector);

	for (i = disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
		/* Only process blocks that are known to be uptodate */
1014
		if (test_bit(R5_Wantdrain, &dev->flags))
1015 1016 1017
			xor_srcs[count++] = dev->page;
	}

D
Dan Williams 已提交
1018
	init_async_submit(&submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
1019
			  ops_complete_prexor, sh, to_addr_conv(sh, percpu));
1020
	tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE, &submit);
1021 1022 1023 1024 1025

	return tx;
}

static struct dma_async_tx_descriptor *
1026
ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
1027 1028
{
	int disks = sh->disks;
1029
	int i;
1030

1031
	pr_debug("%s: stripe %llu\n", __func__,
1032 1033 1034 1035 1036 1037
		(unsigned long long)sh->sector);

	for (i = disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
		struct bio *chosen;

1038
		if (test_and_clear_bit(R5_Wantdrain, &dev->flags)) {
1039 1040
			struct bio *wbi;

1041
			spin_lock_irq(&sh->raid_conf->device_lock);
1042 1043 1044 1045
			chosen = dev->towrite;
			dev->towrite = NULL;
			BUG_ON(dev->written);
			wbi = dev->written = chosen;
1046
			spin_unlock_irq(&sh->raid_conf->device_lock);
1047 1048 1049

			while (wbi && wbi->bi_sector <
				dev->sector + STRIPE_SECTORS) {
T
Tejun Heo 已提交
1050 1051
				if (wbi->bi_rw & REQ_FUA)
					set_bit(R5_WantFUA, &dev->flags);
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
				tx = async_copy_data(1, wbi, dev->page,
					dev->sector, tx);
				wbi = r5_next_bio(wbi, dev->sector);
			}
		}
	}

	return tx;
}

1062
static void ops_complete_reconstruct(void *stripe_head_ref)
1063 1064
{
	struct stripe_head *sh = stripe_head_ref;
1065 1066 1067 1068
	int disks = sh->disks;
	int pd_idx = sh->pd_idx;
	int qd_idx = sh->qd_idx;
	int i;
T
Tejun Heo 已提交
1069
	bool fua = false;
1070

1071
	pr_debug("%s: stripe %llu\n", __func__,
1072 1073
		(unsigned long long)sh->sector);

T
Tejun Heo 已提交
1074 1075 1076
	for (i = disks; i--; )
		fua |= test_bit(R5_WantFUA, &sh->dev[i].flags);

1077 1078
	for (i = disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
1079

T
Tejun Heo 已提交
1080
		if (dev->written || i == pd_idx || i == qd_idx) {
1081
			set_bit(R5_UPTODATE, &dev->flags);
T
Tejun Heo 已提交
1082 1083 1084
			if (fua)
				set_bit(R5_WantFUA, &dev->flags);
		}
1085 1086
	}

1087 1088 1089 1090 1091 1092 1093 1094
	if (sh->reconstruct_state == reconstruct_state_drain_run)
		sh->reconstruct_state = reconstruct_state_drain_result;
	else if (sh->reconstruct_state == reconstruct_state_prexor_drain_run)
		sh->reconstruct_state = reconstruct_state_prexor_drain_result;
	else {
		BUG_ON(sh->reconstruct_state != reconstruct_state_run);
		sh->reconstruct_state = reconstruct_state_result;
	}
1095 1096 1097 1098 1099 1100

	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

static void
1101 1102
ops_run_reconstruct5(struct stripe_head *sh, struct raid5_percpu *percpu,
		     struct dma_async_tx_descriptor *tx)
1103 1104
{
	int disks = sh->disks;
1105
	struct page **xor_srcs = percpu->scribble;
1106
	struct async_submit_ctl submit;
1107 1108
	int count = 0, pd_idx = sh->pd_idx, i;
	struct page *xor_dest;
1109
	int prexor = 0;
1110 1111
	unsigned long flags;

1112
	pr_debug("%s: stripe %llu\n", __func__,
1113 1114 1115 1116 1117
		(unsigned long long)sh->sector);

	/* check if prexor is active which means only process blocks
	 * that are part of a read-modify-write (written)
	 */
1118 1119
	if (sh->reconstruct_state == reconstruct_state_prexor_drain_run) {
		prexor = 1;
1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139
		xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (dev->written)
				xor_srcs[count++] = dev->page;
		}
	} else {
		xor_dest = sh->dev[pd_idx].page;
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (i != pd_idx)
				xor_srcs[count++] = dev->page;
		}
	}

	/* 1/ if we prexor'd then the dest is reused as a source
	 * 2/ if we did not prexor then we are redoing the parity
	 * set ASYNC_TX_XOR_DROP_DST and ASYNC_TX_XOR_ZERO_DST
	 * for the synchronous xor case
	 */
1140
	flags = ASYNC_TX_ACK |
1141 1142 1143 1144
		(prexor ? ASYNC_TX_XOR_DROP_DST : ASYNC_TX_XOR_ZERO_DST);

	atomic_inc(&sh->count);

1145
	init_async_submit(&submit, flags, tx, ops_complete_reconstruct, sh,
1146
			  to_addr_conv(sh, percpu));
1147 1148 1149 1150
	if (unlikely(count == 1))
		tx = async_memcpy(xor_dest, xor_srcs[0], 0, 0, STRIPE_SIZE, &submit);
	else
		tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE, &submit);
1151 1152
}

1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169
static void
ops_run_reconstruct6(struct stripe_head *sh, struct raid5_percpu *percpu,
		     struct dma_async_tx_descriptor *tx)
{
	struct async_submit_ctl submit;
	struct page **blocks = percpu->scribble;
	int count;

	pr_debug("%s: stripe %llu\n", __func__, (unsigned long long)sh->sector);

	count = set_syndrome_sources(blocks, sh);

	atomic_inc(&sh->count);

	init_async_submit(&submit, ASYNC_TX_ACK, tx, ops_complete_reconstruct,
			  sh, to_addr_conv(sh, percpu));
	async_gen_syndrome(blocks, 0, count+2, STRIPE_SIZE,  &submit);
1170 1171 1172 1173 1174 1175
}

static void ops_complete_check(void *stripe_head_ref)
{
	struct stripe_head *sh = stripe_head_ref;

1176
	pr_debug("%s: stripe %llu\n", __func__,
1177 1178
		(unsigned long long)sh->sector);

1179
	sh->check_state = check_state_check_result;
1180 1181 1182 1183
	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

1184
static void ops_run_check_p(struct stripe_head *sh, struct raid5_percpu *percpu)
1185 1186
{
	int disks = sh->disks;
1187 1188 1189
	int pd_idx = sh->pd_idx;
	int qd_idx = sh->qd_idx;
	struct page *xor_dest;
1190
	struct page **xor_srcs = percpu->scribble;
1191
	struct dma_async_tx_descriptor *tx;
1192
	struct async_submit_ctl submit;
1193 1194
	int count;
	int i;
1195

1196
	pr_debug("%s: stripe %llu\n", __func__,
1197 1198
		(unsigned long long)sh->sector);

1199 1200 1201
	count = 0;
	xor_dest = sh->dev[pd_idx].page;
	xor_srcs[count++] = xor_dest;
1202
	for (i = disks; i--; ) {
1203 1204 1205
		if (i == pd_idx || i == qd_idx)
			continue;
		xor_srcs[count++] = sh->dev[i].page;
1206 1207
	}

1208 1209
	init_async_submit(&submit, 0, NULL, NULL, NULL,
			  to_addr_conv(sh, percpu));
D
Dan Williams 已提交
1210
	tx = async_xor_val(xor_dest, xor_srcs, 0, count, STRIPE_SIZE,
1211
			   &sh->ops.zero_sum_result, &submit);
1212 1213

	atomic_inc(&sh->count);
1214 1215
	init_async_submit(&submit, ASYNC_TX_ACK, tx, ops_complete_check, sh, NULL);
	tx = async_trigger_callback(&submit);
1216 1217
}

1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229
static void ops_run_check_pq(struct stripe_head *sh, struct raid5_percpu *percpu, int checkp)
{
	struct page **srcs = percpu->scribble;
	struct async_submit_ctl submit;
	int count;

	pr_debug("%s: stripe %llu checkp: %d\n", __func__,
		(unsigned long long)sh->sector, checkp);

	count = set_syndrome_sources(srcs, sh);
	if (!checkp)
		srcs[count] = NULL;
1230 1231

	atomic_inc(&sh->count);
1232 1233 1234 1235
	init_async_submit(&submit, ASYNC_TX_ACK, NULL, ops_complete_check,
			  sh, to_addr_conv(sh, percpu));
	async_syndrome_val(srcs, 0, count+2, STRIPE_SIZE,
			   &sh->ops.zero_sum_result, percpu->spare_page, &submit);
1236 1237
}

1238
static void __raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
1239 1240 1241
{
	int overlap_clear = 0, i, disks = sh->disks;
	struct dma_async_tx_descriptor *tx = NULL;
1242
	raid5_conf_t *conf = sh->raid_conf;
1243
	int level = conf->level;
1244 1245
	struct raid5_percpu *percpu;
	unsigned long cpu;
1246

1247 1248
	cpu = get_cpu();
	percpu = per_cpu_ptr(conf->percpu, cpu);
1249
	if (test_bit(STRIPE_OP_BIOFILL, &ops_request)) {
1250 1251 1252 1253
		ops_run_biofill(sh);
		overlap_clear++;
	}

1254
	if (test_bit(STRIPE_OP_COMPUTE_BLK, &ops_request)) {
1255 1256 1257 1258 1259 1260 1261 1262 1263 1264
		if (level < 6)
			tx = ops_run_compute5(sh, percpu);
		else {
			if (sh->ops.target2 < 0 || sh->ops.target < 0)
				tx = ops_run_compute6_1(sh, percpu);
			else
				tx = ops_run_compute6_2(sh, percpu);
		}
		/* terminate the chain if reconstruct is not set to be run */
		if (tx && !test_bit(STRIPE_OP_RECONSTRUCT, &ops_request))
1265 1266
			async_tx_ack(tx);
	}
1267

1268
	if (test_bit(STRIPE_OP_PREXOR, &ops_request))
1269
		tx = ops_run_prexor(sh, percpu, tx);
1270

1271
	if (test_bit(STRIPE_OP_BIODRAIN, &ops_request)) {
1272
		tx = ops_run_biodrain(sh, tx);
1273 1274 1275
		overlap_clear++;
	}

1276 1277 1278 1279 1280 1281
	if (test_bit(STRIPE_OP_RECONSTRUCT, &ops_request)) {
		if (level < 6)
			ops_run_reconstruct5(sh, percpu, tx);
		else
			ops_run_reconstruct6(sh, percpu, tx);
	}
1282

1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
	if (test_bit(STRIPE_OP_CHECK, &ops_request)) {
		if (sh->check_state == check_state_run)
			ops_run_check_p(sh, percpu);
		else if (sh->check_state == check_state_run_q)
			ops_run_check_pq(sh, percpu, 0);
		else if (sh->check_state == check_state_run_pq)
			ops_run_check_pq(sh, percpu, 1);
		else
			BUG();
	}
1293 1294 1295 1296 1297 1298 1299

	if (overlap_clear)
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (test_and_clear_bit(R5_Overlap, &dev->flags))
				wake_up(&sh->raid_conf->wait_for_overlap);
		}
1300
	put_cpu();
1301 1302
}

1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332
#ifdef CONFIG_MULTICORE_RAID456
static void async_run_ops(void *param, async_cookie_t cookie)
{
	struct stripe_head *sh = param;
	unsigned long ops_request = sh->ops.request;

	clear_bit_unlock(STRIPE_OPS_REQ_PENDING, &sh->state);
	wake_up(&sh->ops.wait_for_ops);

	__raid_run_ops(sh, ops_request);
	release_stripe(sh);
}

static void raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
{
	/* since handle_stripe can be called outside of raid5d context
	 * we need to ensure sh->ops.request is de-staged before another
	 * request arrives
	 */
	wait_event(sh->ops.wait_for_ops,
		   !test_and_set_bit_lock(STRIPE_OPS_REQ_PENDING, &sh->state));
	sh->ops.request = ops_request;

	atomic_inc(&sh->count);
	async_schedule(async_run_ops, sh);
}
#else
#define raid_run_ops __raid_run_ops
#endif

1333
static int grow_one_stripe(raid5_conf_t *conf)
L
Linus Torvalds 已提交
1334 1335
{
	struct stripe_head *sh;
N
Namhyung Kim 已提交
1336
	sh = kmem_cache_zalloc(conf->slab_cache, GFP_KERNEL);
1337 1338
	if (!sh)
		return 0;
N
Namhyung Kim 已提交
1339

1340
	sh->raid_conf = conf;
1341 1342 1343
	#ifdef CONFIG_MULTICORE_RAID456
	init_waitqueue_head(&sh->ops.wait_for_ops);
	#endif
1344

1345 1346
	if (grow_buffers(sh)) {
		shrink_buffers(sh);
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
		kmem_cache_free(conf->slab_cache, sh);
		return 0;
	}
	/* we just created an active stripe so... */
	atomic_set(&sh->count, 1);
	atomic_inc(&conf->active_stripes);
	INIT_LIST_HEAD(&sh->lru);
	release_stripe(sh);
	return 1;
}

static int grow_stripes(raid5_conf_t *conf, int num)
{
1360
	struct kmem_cache *sc;
1361
	int devs = max(conf->raid_disks, conf->previous_raid_disks);
L
Linus Torvalds 已提交
1362

1363 1364 1365 1366 1367 1368 1369 1370
	if (conf->mddev->gendisk)
		sprintf(conf->cache_name[0],
			"raid%d-%s", conf->level, mdname(conf->mddev));
	else
		sprintf(conf->cache_name[0],
			"raid%d-%p", conf->level, conf->mddev);
	sprintf(conf->cache_name[1], "%s-alt", conf->cache_name[0]);

1371 1372
	conf->active_name = 0;
	sc = kmem_cache_create(conf->cache_name[conf->active_name],
L
Linus Torvalds 已提交
1373
			       sizeof(struct stripe_head)+(devs-1)*sizeof(struct r5dev),
1374
			       0, 0, NULL);
L
Linus Torvalds 已提交
1375 1376 1377
	if (!sc)
		return 1;
	conf->slab_cache = sc;
1378
	conf->pool_size = devs;
1379
	while (num--)
1380
		if (!grow_one_stripe(conf))
L
Linus Torvalds 已提交
1381 1382 1383
			return 1;
	return 0;
}
1384

1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
/**
 * scribble_len - return the required size of the scribble region
 * @num - total number of disks in the array
 *
 * The size must be enough to contain:
 * 1/ a struct page pointer for each device in the array +2
 * 2/ room to convert each entry in (1) to its corresponding dma
 *    (dma_map_page()) or page (page_address()) address.
 *
 * Note: the +2 is for the destination buffers of the ddf/raid6 case where we
 * calculate over all devices (not just the data blocks), using zeros in place
 * of the P and Q blocks.
 */
static size_t scribble_len(int num)
{
	size_t len;

	len = sizeof(struct page *) * (num+2) + sizeof(addr_conv_t) * (num+2);

	return len;
}

1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
static int resize_stripes(raid5_conf_t *conf, int newsize)
{
	/* Make all the stripes able to hold 'newsize' devices.
	 * New slots in each stripe get 'page' set to a new page.
	 *
	 * This happens in stages:
	 * 1/ create a new kmem_cache and allocate the required number of
	 *    stripe_heads.
	 * 2/ gather all the old stripe_heads and tranfer the pages across
	 *    to the new stripe_heads.  This will have the side effect of
	 *    freezing the array as once all stripe_heads have been collected,
	 *    no IO will be possible.  Old stripe heads are freed once their
	 *    pages have been transferred over, and the old kmem_cache is
	 *    freed when all stripes are done.
	 * 3/ reallocate conf->disks to be suitable bigger.  If this fails,
	 *    we simple return a failre status - no need to clean anything up.
	 * 4/ allocate new pages for the new slots in the new stripe_heads.
	 *    If this fails, we don't bother trying the shrink the
	 *    stripe_heads down again, we just leave them as they are.
	 *    As each stripe_head is processed the new one is released into
	 *    active service.
	 *
	 * Once step2 is started, we cannot afford to wait for a write,
	 * so we use GFP_NOIO allocations.
	 */
	struct stripe_head *osh, *nsh;
	LIST_HEAD(newstripes);
	struct disk_info *ndisks;
1435
	unsigned long cpu;
1436
	int err;
1437
	struct kmem_cache *sc;
1438 1439 1440 1441 1442
	int i;

	if (newsize <= conf->pool_size)
		return 0; /* never bother to shrink */

1443 1444 1445
	err = md_allow_write(conf->mddev);
	if (err)
		return err;
1446

1447 1448 1449
	/* Step 1 */
	sc = kmem_cache_create(conf->cache_name[1-conf->active_name],
			       sizeof(struct stripe_head)+(newsize-1)*sizeof(struct r5dev),
1450
			       0, 0, NULL);
1451 1452 1453 1454
	if (!sc)
		return -ENOMEM;

	for (i = conf->max_nr_stripes; i; i--) {
N
Namhyung Kim 已提交
1455
		nsh = kmem_cache_zalloc(sc, GFP_KERNEL);
1456 1457 1458 1459
		if (!nsh)
			break;

		nsh->raid_conf = conf;
1460 1461 1462
		#ifdef CONFIG_MULTICORE_RAID456
		init_waitqueue_head(&nsh->ops.wait_for_ops);
		#endif
1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484

		list_add(&nsh->lru, &newstripes);
	}
	if (i) {
		/* didn't get enough, give up */
		while (!list_empty(&newstripes)) {
			nsh = list_entry(newstripes.next, struct stripe_head, lru);
			list_del(&nsh->lru);
			kmem_cache_free(sc, nsh);
		}
		kmem_cache_destroy(sc);
		return -ENOMEM;
	}
	/* Step 2 - Must use GFP_NOIO now.
	 * OK, we have enough stripes, start collecting inactive
	 * stripes and copying them over
	 */
	list_for_each_entry(nsh, &newstripes, lru) {
		spin_lock_irq(&conf->device_lock);
		wait_event_lock_irq(conf->wait_for_stripe,
				    !list_empty(&conf->inactive_list),
				    conf->device_lock,
N
NeilBrown 已提交
1485
				    );
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499
		osh = get_free_stripe(conf);
		spin_unlock_irq(&conf->device_lock);
		atomic_set(&nsh->count, 1);
		for(i=0; i<conf->pool_size; i++)
			nsh->dev[i].page = osh->dev[i].page;
		for( ; i<newsize; i++)
			nsh->dev[i].page = NULL;
		kmem_cache_free(conf->slab_cache, osh);
	}
	kmem_cache_destroy(conf->slab_cache);

	/* Step 3.
	 * At this point, we are holding all the stripes so the array
	 * is completely stalled, so now is a good time to resize
1500
	 * conf->disks and the scribble region
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
	 */
	ndisks = kzalloc(newsize * sizeof(struct disk_info), GFP_NOIO);
	if (ndisks) {
		for (i=0; i<conf->raid_disks; i++)
			ndisks[i] = conf->disks[i];
		kfree(conf->disks);
		conf->disks = ndisks;
	} else
		err = -ENOMEM;

1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529
	get_online_cpus();
	conf->scribble_len = scribble_len(newsize);
	for_each_present_cpu(cpu) {
		struct raid5_percpu *percpu;
		void *scribble;

		percpu = per_cpu_ptr(conf->percpu, cpu);
		scribble = kmalloc(conf->scribble_len, GFP_NOIO);

		if (scribble) {
			kfree(percpu->scribble);
			percpu->scribble = scribble;
		} else {
			err = -ENOMEM;
			break;
		}
	}
	put_online_cpus();

1530 1531 1532 1533
	/* Step 4, return new stripes to service */
	while(!list_empty(&newstripes)) {
		nsh = list_entry(newstripes.next, struct stripe_head, lru);
		list_del_init(&nsh->lru);
1534

1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
		for (i=conf->raid_disks; i < newsize; i++)
			if (nsh->dev[i].page == NULL) {
				struct page *p = alloc_page(GFP_NOIO);
				nsh->dev[i].page = p;
				if (!p)
					err = -ENOMEM;
			}
		release_stripe(nsh);
	}
	/* critical section pass, GFP_NOIO no longer needed */

	conf->slab_cache = sc;
	conf->active_name = 1-conf->active_name;
	conf->pool_size = newsize;
	return err;
}
L
Linus Torvalds 已提交
1551

1552
static int drop_one_stripe(raid5_conf_t *conf)
L
Linus Torvalds 已提交
1553 1554 1555
{
	struct stripe_head *sh;

1556 1557 1558 1559 1560
	spin_lock_irq(&conf->device_lock);
	sh = get_free_stripe(conf);
	spin_unlock_irq(&conf->device_lock);
	if (!sh)
		return 0;
1561
	BUG_ON(atomic_read(&sh->count));
1562
	shrink_buffers(sh);
1563 1564 1565 1566 1567 1568 1569 1570 1571 1572
	kmem_cache_free(conf->slab_cache, sh);
	atomic_dec(&conf->active_stripes);
	return 1;
}

static void shrink_stripes(raid5_conf_t *conf)
{
	while (drop_one_stripe(conf))
		;

N
NeilBrown 已提交
1573 1574
	if (conf->slab_cache)
		kmem_cache_destroy(conf->slab_cache);
L
Linus Torvalds 已提交
1575 1576 1577
	conf->slab_cache = NULL;
}

1578
static void raid5_end_read_request(struct bio * bi, int error)
L
Linus Torvalds 已提交
1579
{
1580
	struct stripe_head *sh = bi->bi_private;
L
Linus Torvalds 已提交
1581
	raid5_conf_t *conf = sh->raid_conf;
1582
	int disks = sh->disks, i;
L
Linus Torvalds 已提交
1583
	int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags);
1584 1585
	char b[BDEVNAME_SIZE];
	mdk_rdev_t *rdev;
L
Linus Torvalds 已提交
1586 1587 1588 1589 1590 1591


	for (i=0 ; i<disks; i++)
		if (bi == &sh->dev[i].req)
			break;

1592 1593
	pr_debug("end_read_request %llu/%d, count: %d, uptodate %d.\n",
		(unsigned long long)sh->sector, i, atomic_read(&sh->count),
L
Linus Torvalds 已提交
1594 1595 1596
		uptodate);
	if (i == disks) {
		BUG();
1597
		return;
L
Linus Torvalds 已提交
1598 1599 1600 1601
	}

	if (uptodate) {
		set_bit(R5_UPTODATE, &sh->dev[i].flags);
1602
		if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
1603
			rdev = conf->disks[i].rdev;
1604 1605 1606 1607 1608 1609 1610 1611
			printk_ratelimited(
				KERN_INFO
				"md/raid:%s: read error corrected"
				" (%lu sectors at %llu on %s)\n",
				mdname(conf->mddev), STRIPE_SECTORS,
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdevname(rdev->bdev, b));
1612
			atomic_add(STRIPE_SECTORS, &rdev->corrected_errors);
1613 1614 1615
			clear_bit(R5_ReadError, &sh->dev[i].flags);
			clear_bit(R5_ReWrite, &sh->dev[i].flags);
		}
1616 1617
		if (atomic_read(&conf->disks[i].rdev->read_errors))
			atomic_set(&conf->disks[i].rdev->read_errors, 0);
L
Linus Torvalds 已提交
1618
	} else {
1619
		const char *bdn = bdevname(conf->disks[i].rdev->bdev, b);
1620
		int retry = 0;
1621 1622
		rdev = conf->disks[i].rdev;

L
Linus Torvalds 已提交
1623
		clear_bit(R5_UPTODATE, &sh->dev[i].flags);
1624
		atomic_inc(&rdev->read_errors);
1625
		if (conf->mddev->degraded >= conf->max_degraded)
1626 1627 1628 1629 1630 1631 1632 1633
			printk_ratelimited(
				KERN_WARNING
				"md/raid:%s: read error not correctable "
				"(sector %llu on %s).\n",
				mdname(conf->mddev),
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdn);
1634
		else if (test_bit(R5_ReWrite, &sh->dev[i].flags))
1635
			/* Oh, no!!! */
1636 1637 1638 1639 1640 1641 1642 1643
			printk_ratelimited(
				KERN_WARNING
				"md/raid:%s: read error NOT corrected!! "
				"(sector %llu on %s).\n",
				mdname(conf->mddev),
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdn);
1644
		else if (atomic_read(&rdev->read_errors)
1645
			 > conf->max_nr_stripes)
N
NeilBrown 已提交
1646
			printk(KERN_WARNING
1647
			       "md/raid:%s: Too many read errors, failing device %s.\n",
1648
			       mdname(conf->mddev), bdn);
1649 1650 1651 1652 1653
		else
			retry = 1;
		if (retry)
			set_bit(R5_ReadError, &sh->dev[i].flags);
		else {
1654 1655
			clear_bit(R5_ReadError, &sh->dev[i].flags);
			clear_bit(R5_ReWrite, &sh->dev[i].flags);
1656
			md_error(conf->mddev, rdev);
1657
		}
L
Linus Torvalds 已提交
1658 1659 1660 1661 1662 1663 1664
	}
	rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
	clear_bit(R5_LOCKED, &sh->dev[i].flags);
	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

1665
static void raid5_end_write_request(struct bio *bi, int error)
L
Linus Torvalds 已提交
1666
{
1667
	struct stripe_head *sh = bi->bi_private;
L
Linus Torvalds 已提交
1668
	raid5_conf_t *conf = sh->raid_conf;
1669
	int disks = sh->disks, i;
L
Linus Torvalds 已提交
1670
	int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags);
1671 1672
	sector_t first_bad;
	int bad_sectors;
L
Linus Torvalds 已提交
1673 1674 1675 1676 1677

	for (i=0 ; i<disks; i++)
		if (bi == &sh->dev[i].req)
			break;

1678
	pr_debug("end_write_request %llu/%d, count %d, uptodate: %d.\n",
L
Linus Torvalds 已提交
1679 1680 1681 1682
		(unsigned long long)sh->sector, i, atomic_read(&sh->count),
		uptodate);
	if (i == disks) {
		BUG();
1683
		return;
L
Linus Torvalds 已提交
1684 1685
	}

1686 1687 1688
	if (!uptodate) {
		set_bit(WriteErrorSeen, &conf->disks[i].rdev->flags);
		set_bit(R5_WriteError, &sh->dev[i].flags);
1689 1690 1691
	} else if (is_badblock(conf->disks[i].rdev, sh->sector, STRIPE_SECTORS,
			       &first_bad, &bad_sectors))
		set_bit(R5_MadeGood, &sh->dev[i].flags);
L
Linus Torvalds 已提交
1692 1693 1694 1695 1696

	rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
	
	clear_bit(R5_LOCKED, &sh->dev[i].flags);
	set_bit(STRIPE_HANDLE, &sh->state);
1697
	release_stripe(sh);
L
Linus Torvalds 已提交
1698 1699 1700
}


1701
static sector_t compute_blocknr(struct stripe_head *sh, int i, int previous);
L
Linus Torvalds 已提交
1702
	
1703
static void raid5_build_block(struct stripe_head *sh, int i, int previous)
L
Linus Torvalds 已提交
1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
{
	struct r5dev *dev = &sh->dev[i];

	bio_init(&dev->req);
	dev->req.bi_io_vec = &dev->vec;
	dev->req.bi_vcnt++;
	dev->req.bi_max_vecs++;
	dev->vec.bv_page = dev->page;
	dev->vec.bv_len = STRIPE_SIZE;
	dev->vec.bv_offset = 0;

	dev->req.bi_sector = sh->sector;
	dev->req.bi_private = sh;

	dev->flags = 0;
1719
	dev->sector = compute_blocknr(sh, i, previous);
L
Linus Torvalds 已提交
1720 1721 1722 1723 1724
}

static void error(mddev_t *mddev, mdk_rdev_t *rdev)
{
	char b[BDEVNAME_SIZE];
1725
	raid5_conf_t *conf = mddev->private;
1726
	pr_debug("raid456: error called\n");
L
Linus Torvalds 已提交
1727

1728 1729 1730 1731 1732 1733 1734 1735 1736
	if (test_and_clear_bit(In_sync, &rdev->flags)) {
		unsigned long flags;
		spin_lock_irqsave(&conf->device_lock, flags);
		mddev->degraded++;
		spin_unlock_irqrestore(&conf->device_lock, flags);
		/*
		 * if recovery was running, make sure it aborts.
		 */
		set_bit(MD_RECOVERY_INTR, &mddev->recovery);
L
Linus Torvalds 已提交
1737
	}
1738
	set_bit(Blocked, &rdev->flags);
1739 1740 1741 1742 1743 1744 1745 1746 1747
	set_bit(Faulty, &rdev->flags);
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
	printk(KERN_ALERT
	       "md/raid:%s: Disk failure on %s, disabling device.\n"
	       "md/raid:%s: Operation continuing on %d devices.\n",
	       mdname(mddev),
	       bdevname(rdev->bdev, b),
	       mdname(mddev),
	       conf->raid_disks - mddev->degraded);
1748
}
L
Linus Torvalds 已提交
1749 1750 1751 1752 1753

/*
 * Input: a 'big' sector number,
 * Output: index of the data and parity disk, and the sector # in them.
 */
1754
static sector_t raid5_compute_sector(raid5_conf_t *conf, sector_t r_sector,
1755 1756
				     int previous, int *dd_idx,
				     struct stripe_head *sh)
L
Linus Torvalds 已提交
1757
{
N
NeilBrown 已提交
1758
	sector_t stripe, stripe2;
1759
	sector_t chunk_number;
L
Linus Torvalds 已提交
1760
	unsigned int chunk_offset;
1761
	int pd_idx, qd_idx;
1762
	int ddf_layout = 0;
L
Linus Torvalds 已提交
1763
	sector_t new_sector;
1764 1765
	int algorithm = previous ? conf->prev_algo
				 : conf->algorithm;
1766 1767
	int sectors_per_chunk = previous ? conf->prev_chunk_sectors
					 : conf->chunk_sectors;
1768 1769 1770
	int raid_disks = previous ? conf->previous_raid_disks
				  : conf->raid_disks;
	int data_disks = raid_disks - conf->max_degraded;
L
Linus Torvalds 已提交
1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782

	/* First compute the information on this sector */

	/*
	 * Compute the chunk number and the sector offset inside the chunk
	 */
	chunk_offset = sector_div(r_sector, sectors_per_chunk);
	chunk_number = r_sector;

	/*
	 * Compute the stripe number
	 */
1783 1784
	stripe = chunk_number;
	*dd_idx = sector_div(stripe, data_disks);
N
NeilBrown 已提交
1785
	stripe2 = stripe;
L
Linus Torvalds 已提交
1786 1787 1788
	/*
	 * Select the parity disk based on the user selected algorithm.
	 */
1789
	pd_idx = qd_idx = -1;
1790 1791
	switch(conf->level) {
	case 4:
1792
		pd_idx = data_disks;
1793 1794
		break;
	case 5:
1795
		switch (algorithm) {
L
Linus Torvalds 已提交
1796
		case ALGORITHM_LEFT_ASYMMETRIC:
N
NeilBrown 已提交
1797
			pd_idx = data_disks - sector_div(stripe2, raid_disks);
1798
			if (*dd_idx >= pd_idx)
L
Linus Torvalds 已提交
1799 1800 1801
				(*dd_idx)++;
			break;
		case ALGORITHM_RIGHT_ASYMMETRIC:
N
NeilBrown 已提交
1802
			pd_idx = sector_div(stripe2, raid_disks);
1803
			if (*dd_idx >= pd_idx)
L
Linus Torvalds 已提交
1804 1805 1806
				(*dd_idx)++;
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
N
NeilBrown 已提交
1807
			pd_idx = data_disks - sector_div(stripe2, raid_disks);
1808
			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
L
Linus Torvalds 已提交
1809 1810
			break;
		case ALGORITHM_RIGHT_SYMMETRIC:
N
NeilBrown 已提交
1811
			pd_idx = sector_div(stripe2, raid_disks);
1812
			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
L
Linus Torvalds 已提交
1813
			break;
1814 1815 1816 1817 1818 1819 1820
		case ALGORITHM_PARITY_0:
			pd_idx = 0;
			(*dd_idx)++;
			break;
		case ALGORITHM_PARITY_N:
			pd_idx = data_disks;
			break;
L
Linus Torvalds 已提交
1821
		default:
1822
			BUG();
1823 1824 1825 1826
		}
		break;
	case 6:

1827
		switch (algorithm) {
1828
		case ALGORITHM_LEFT_ASYMMETRIC:
N
NeilBrown 已提交
1829
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1830 1831
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
1832
				(*dd_idx)++;	/* Q D D D P */
1833 1834
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
1835 1836 1837
				(*dd_idx) += 2; /* D D P Q D */
			break;
		case ALGORITHM_RIGHT_ASYMMETRIC:
N
NeilBrown 已提交
1838
			pd_idx = sector_div(stripe2, raid_disks);
1839 1840
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
1841
				(*dd_idx)++;	/* Q D D D P */
1842 1843
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
1844 1845 1846
				(*dd_idx) += 2; /* D D P Q D */
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
N
NeilBrown 已提交
1847
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1848 1849
			qd_idx = (pd_idx + 1) % raid_disks;
			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
1850 1851
			break;
		case ALGORITHM_RIGHT_SYMMETRIC:
N
NeilBrown 已提交
1852
			pd_idx = sector_div(stripe2, raid_disks);
1853 1854
			qd_idx = (pd_idx + 1) % raid_disks;
			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
1855
			break;
1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870

		case ALGORITHM_PARITY_0:
			pd_idx = 0;
			qd_idx = 1;
			(*dd_idx) += 2;
			break;
		case ALGORITHM_PARITY_N:
			pd_idx = data_disks;
			qd_idx = data_disks + 1;
			break;

		case ALGORITHM_ROTATING_ZERO_RESTART:
			/* Exactly the same as RIGHT_ASYMMETRIC, but or
			 * of blocks for computing Q is different.
			 */
N
NeilBrown 已提交
1871
			pd_idx = sector_div(stripe2, raid_disks);
1872 1873 1874 1875 1876 1877
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
				(*dd_idx)++;	/* Q D D D P */
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
				(*dd_idx) += 2; /* D D P Q D */
1878
			ddf_layout = 1;
1879 1880 1881 1882 1883 1884 1885
			break;

		case ALGORITHM_ROTATING_N_RESTART:
			/* Same a left_asymmetric, by first stripe is
			 * D D D P Q  rather than
			 * Q D D D P
			 */
N
NeilBrown 已提交
1886 1887
			stripe2 += 1;
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1888 1889 1890 1891 1892 1893
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
				(*dd_idx)++;	/* Q D D D P */
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
				(*dd_idx) += 2; /* D D P Q D */
1894
			ddf_layout = 1;
1895 1896 1897 1898
			break;

		case ALGORITHM_ROTATING_N_CONTINUE:
			/* Same as left_symmetric but Q is before P */
N
NeilBrown 已提交
1899
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1900 1901
			qd_idx = (pd_idx + raid_disks - 1) % raid_disks;
			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
1902
			ddf_layout = 1;
1903 1904 1905 1906
			break;

		case ALGORITHM_LEFT_ASYMMETRIC_6:
			/* RAID5 left_asymmetric, with Q on last device */
N
NeilBrown 已提交
1907
			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
1908 1909 1910 1911 1912 1913
			if (*dd_idx >= pd_idx)
				(*dd_idx)++;
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_RIGHT_ASYMMETRIC_6:
N
NeilBrown 已提交
1914
			pd_idx = sector_div(stripe2, raid_disks-1);
1915 1916 1917 1918 1919 1920
			if (*dd_idx >= pd_idx)
				(*dd_idx)++;
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_LEFT_SYMMETRIC_6:
N
NeilBrown 已提交
1921
			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
1922 1923 1924 1925 1926
			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_RIGHT_SYMMETRIC_6:
N
NeilBrown 已提交
1927
			pd_idx = sector_div(stripe2, raid_disks-1);
1928 1929 1930 1931 1932 1933 1934 1935 1936 1937
			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_PARITY_0_6:
			pd_idx = 0;
			(*dd_idx)++;
			qd_idx = raid_disks - 1;
			break;

1938
		default:
1939
			BUG();
1940 1941
		}
		break;
L
Linus Torvalds 已提交
1942 1943
	}

1944 1945 1946
	if (sh) {
		sh->pd_idx = pd_idx;
		sh->qd_idx = qd_idx;
1947
		sh->ddf_layout = ddf_layout;
1948
	}
L
Linus Torvalds 已提交
1949 1950 1951 1952 1953 1954 1955 1956
	/*
	 * Finally, compute the new sector number
	 */
	new_sector = (sector_t)stripe * sectors_per_chunk + chunk_offset;
	return new_sector;
}


1957
static sector_t compute_blocknr(struct stripe_head *sh, int i, int previous)
L
Linus Torvalds 已提交
1958 1959
{
	raid5_conf_t *conf = sh->raid_conf;
1960 1961
	int raid_disks = sh->disks;
	int data_disks = raid_disks - conf->max_degraded;
L
Linus Torvalds 已提交
1962
	sector_t new_sector = sh->sector, check;
1963 1964
	int sectors_per_chunk = previous ? conf->prev_chunk_sectors
					 : conf->chunk_sectors;
1965 1966
	int algorithm = previous ? conf->prev_algo
				 : conf->algorithm;
L
Linus Torvalds 已提交
1967 1968
	sector_t stripe;
	int chunk_offset;
1969 1970
	sector_t chunk_number;
	int dummy1, dd_idx = i;
L
Linus Torvalds 已提交
1971
	sector_t r_sector;
1972
	struct stripe_head sh2;
L
Linus Torvalds 已提交
1973

1974

L
Linus Torvalds 已提交
1975 1976 1977
	chunk_offset = sector_div(new_sector, sectors_per_chunk);
	stripe = new_sector;

1978 1979 1980 1981 1982
	if (i == sh->pd_idx)
		return 0;
	switch(conf->level) {
	case 4: break;
	case 5:
1983
		switch (algorithm) {
L
Linus Torvalds 已提交
1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
		case ALGORITHM_LEFT_ASYMMETRIC:
		case ALGORITHM_RIGHT_ASYMMETRIC:
			if (i > sh->pd_idx)
				i--;
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
		case ALGORITHM_RIGHT_SYMMETRIC:
			if (i < sh->pd_idx)
				i += raid_disks;
			i -= (sh->pd_idx + 1);
			break;
1995 1996 1997 1998 1999
		case ALGORITHM_PARITY_0:
			i -= 1;
			break;
		case ALGORITHM_PARITY_N:
			break;
L
Linus Torvalds 已提交
2000
		default:
2001
			BUG();
2002 2003 2004
		}
		break;
	case 6:
2005
		if (i == sh->qd_idx)
2006
			return 0; /* It is the Q disk */
2007
		switch (algorithm) {
2008 2009
		case ALGORITHM_LEFT_ASYMMETRIC:
		case ALGORITHM_RIGHT_ASYMMETRIC:
2010 2011 2012 2013
		case ALGORITHM_ROTATING_ZERO_RESTART:
		case ALGORITHM_ROTATING_N_RESTART:
			if (sh->pd_idx == raid_disks-1)
				i--;	/* Q D D D P */
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
			else if (i > sh->pd_idx)
				i -= 2; /* D D P Q D */
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
		case ALGORITHM_RIGHT_SYMMETRIC:
			if (sh->pd_idx == raid_disks-1)
				i--; /* Q D D D P */
			else {
				/* D D P Q D */
				if (i < sh->pd_idx)
					i += raid_disks;
				i -= (sh->pd_idx + 2);
			}
			break;
2028 2029 2030 2031 2032 2033
		case ALGORITHM_PARITY_0:
			i -= 2;
			break;
		case ALGORITHM_PARITY_N:
			break;
		case ALGORITHM_ROTATING_N_CONTINUE:
2034
			/* Like left_symmetric, but P is before Q */
2035 2036
			if (sh->pd_idx == 0)
				i--;	/* P D D D Q */
2037 2038 2039 2040 2041 2042
			else {
				/* D D Q P D */
				if (i < sh->pd_idx)
					i += raid_disks;
				i -= (sh->pd_idx + 1);
			}
2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057
			break;
		case ALGORITHM_LEFT_ASYMMETRIC_6:
		case ALGORITHM_RIGHT_ASYMMETRIC_6:
			if (i > sh->pd_idx)
				i--;
			break;
		case ALGORITHM_LEFT_SYMMETRIC_6:
		case ALGORITHM_RIGHT_SYMMETRIC_6:
			if (i < sh->pd_idx)
				i += data_disks + 1;
			i -= (sh->pd_idx + 1);
			break;
		case ALGORITHM_PARITY_0_6:
			i -= 1;
			break;
2058
		default:
2059
			BUG();
2060 2061
		}
		break;
L
Linus Torvalds 已提交
2062 2063 2064
	}

	chunk_number = stripe * data_disks + i;
2065
	r_sector = chunk_number * sectors_per_chunk + chunk_offset;
L
Linus Torvalds 已提交
2066

2067
	check = raid5_compute_sector(conf, r_sector,
2068
				     previous, &dummy1, &sh2);
2069 2070
	if (check != sh->sector || dummy1 != dd_idx || sh2.pd_idx != sh->pd_idx
		|| sh2.qd_idx != sh->qd_idx) {
2071 2072
		printk(KERN_ERR "md/raid:%s: compute_blocknr: map not correct\n",
		       mdname(conf->mddev));
L
Linus Torvalds 已提交
2073 2074 2075 2076 2077 2078
		return 0;
	}
	return r_sector;
}


2079
static void
2080
schedule_reconstruction(struct stripe_head *sh, struct stripe_head_state *s,
2081
			 int rcw, int expand)
2082 2083
{
	int i, pd_idx = sh->pd_idx, disks = sh->disks;
2084 2085
	raid5_conf_t *conf = sh->raid_conf;
	int level = conf->level;
2086 2087 2088 2089 2090 2091 2092

	if (rcw) {
		/* if we are not expanding this is a proper write request, and
		 * there will be bios with new data to be drained into the
		 * stripe cache
		 */
		if (!expand) {
2093 2094 2095 2096
			sh->reconstruct_state = reconstruct_state_drain_run;
			set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
		} else
			sh->reconstruct_state = reconstruct_state_run;
2097

2098
		set_bit(STRIPE_OP_RECONSTRUCT, &s->ops_request);
2099 2100 2101 2102 2103 2104

		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];

			if (dev->towrite) {
				set_bit(R5_LOCKED, &dev->flags);
2105
				set_bit(R5_Wantdrain, &dev->flags);
2106 2107
				if (!expand)
					clear_bit(R5_UPTODATE, &dev->flags);
2108
				s->locked++;
2109 2110
			}
		}
2111
		if (s->locked + conf->max_degraded == disks)
2112
			if (!test_and_set_bit(STRIPE_FULL_WRITE, &sh->state))
2113
				atomic_inc(&conf->pending_full_writes);
2114
	} else {
2115
		BUG_ON(level == 6);
2116 2117 2118
		BUG_ON(!(test_bit(R5_UPTODATE, &sh->dev[pd_idx].flags) ||
			test_bit(R5_Wantcompute, &sh->dev[pd_idx].flags)));

2119
		sh->reconstruct_state = reconstruct_state_prexor_drain_run;
2120 2121
		set_bit(STRIPE_OP_PREXOR, &s->ops_request);
		set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
2122
		set_bit(STRIPE_OP_RECONSTRUCT, &s->ops_request);
2123 2124 2125 2126 2127 2128 2129 2130

		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (i == pd_idx)
				continue;

			if (dev->towrite &&
			    (test_bit(R5_UPTODATE, &dev->flags) ||
2131 2132
			     test_bit(R5_Wantcompute, &dev->flags))) {
				set_bit(R5_Wantdrain, &dev->flags);
2133 2134
				set_bit(R5_LOCKED, &dev->flags);
				clear_bit(R5_UPTODATE, &dev->flags);
2135
				s->locked++;
2136 2137 2138 2139
			}
		}
	}

2140
	/* keep the parity disk(s) locked while asynchronous operations
2141 2142 2143 2144
	 * are in flight
	 */
	set_bit(R5_LOCKED, &sh->dev[pd_idx].flags);
	clear_bit(R5_UPTODATE, &sh->dev[pd_idx].flags);
2145
	s->locked++;
2146

2147 2148 2149 2150 2151 2152 2153 2154 2155
	if (level == 6) {
		int qd_idx = sh->qd_idx;
		struct r5dev *dev = &sh->dev[qd_idx];

		set_bit(R5_LOCKED, &dev->flags);
		clear_bit(R5_UPTODATE, &dev->flags);
		s->locked++;
	}

2156
	pr_debug("%s: stripe %llu locked: %d ops_request: %lx\n",
2157
		__func__, (unsigned long long)sh->sector,
2158
		s->locked, s->ops_request);
2159
}
2160

L
Linus Torvalds 已提交
2161 2162
/*
 * Each stripe/dev can have one or more bion attached.
2163
 * toread/towrite point to the first in a chain.
L
Linus Torvalds 已提交
2164 2165 2166 2167 2168 2169
 * The bi_next chain must be in order.
 */
static int add_stripe_bio(struct stripe_head *sh, struct bio *bi, int dd_idx, int forwrite)
{
	struct bio **bip;
	raid5_conf_t *conf = sh->raid_conf;
2170
	int firstwrite=0;
L
Linus Torvalds 已提交
2171

2172
	pr_debug("adding bi b#%llu to stripe s#%llu\n",
L
Linus Torvalds 已提交
2173 2174 2175 2176 2177
		(unsigned long long)bi->bi_sector,
		(unsigned long long)sh->sector);


	spin_lock_irq(&conf->device_lock);
2178
	if (forwrite) {
L
Linus Torvalds 已提交
2179
		bip = &sh->dev[dd_idx].towrite;
2180 2181 2182
		if (*bip == NULL && sh->dev[dd_idx].written == NULL)
			firstwrite = 1;
	} else
L
Linus Torvalds 已提交
2183 2184 2185 2186 2187 2188 2189 2190 2191
		bip = &sh->dev[dd_idx].toread;
	while (*bip && (*bip)->bi_sector < bi->bi_sector) {
		if ((*bip)->bi_sector + ((*bip)->bi_size >> 9) > bi->bi_sector)
			goto overlap;
		bip = & (*bip)->bi_next;
	}
	if (*bip && (*bip)->bi_sector < bi->bi_sector + ((bi->bi_size)>>9))
		goto overlap;

2192
	BUG_ON(*bip && bi->bi_next && (*bip) != bi->bi_next);
L
Linus Torvalds 已提交
2193 2194 2195
	if (*bip)
		bi->bi_next = *bip;
	*bip = bi;
2196
	bi->bi_phys_segments++;
2197

L
Linus Torvalds 已提交
2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210
	if (forwrite) {
		/* check if page is covered */
		sector_t sector = sh->dev[dd_idx].sector;
		for (bi=sh->dev[dd_idx].towrite;
		     sector < sh->dev[dd_idx].sector + STRIPE_SECTORS &&
			     bi && bi->bi_sector <= sector;
		     bi = r5_next_bio(bi, sh->dev[dd_idx].sector)) {
			if (bi->bi_sector + (bi->bi_size>>9) >= sector)
				sector = bi->bi_sector + (bi->bi_size>>9);
		}
		if (sector >= sh->dev[dd_idx].sector + STRIPE_SECTORS)
			set_bit(R5_OVERWRITE, &sh->dev[dd_idx].flags);
	}
2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222
	spin_unlock_irq(&conf->device_lock);

	pr_debug("added bi b#%llu to stripe s#%llu, disk %d.\n",
		(unsigned long long)(*bip)->bi_sector,
		(unsigned long long)sh->sector, dd_idx);

	if (conf->mddev->bitmap && firstwrite) {
		bitmap_startwrite(conf->mddev->bitmap, sh->sector,
				  STRIPE_SECTORS, 0);
		sh->bm_seq = conf->seq_flush+1;
		set_bit(STRIPE_BIT_DELAY, &sh->state);
	}
L
Linus Torvalds 已提交
2223 2224 2225 2226 2227 2228 2229 2230
	return 1;

 overlap:
	set_bit(R5_Overlap, &sh->dev[dd_idx].flags);
	spin_unlock_irq(&conf->device_lock);
	return 0;
}

2231 2232
static void end_reshape(raid5_conf_t *conf);

2233 2234
static void stripe_set_idx(sector_t stripe, raid5_conf_t *conf, int previous,
			    struct stripe_head *sh)
2235
{
2236
	int sectors_per_chunk =
2237
		previous ? conf->prev_chunk_sectors : conf->chunk_sectors;
2238
	int dd_idx;
2239
	int chunk_offset = sector_div(stripe, sectors_per_chunk);
2240
	int disks = previous ? conf->previous_raid_disks : conf->raid_disks;
2241

2242 2243
	raid5_compute_sector(conf,
			     stripe * (disks - conf->max_degraded)
2244
			     *sectors_per_chunk + chunk_offset,
2245
			     previous,
2246
			     &dd_idx, sh);
2247 2248
}

2249
static void
2250
handle_failed_stripe(raid5_conf_t *conf, struct stripe_head *sh,
2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263
				struct stripe_head_state *s, int disks,
				struct bio **return_bi)
{
	int i;
	for (i = disks; i--; ) {
		struct bio *bi;
		int bitmap_end = 0;

		if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
			mdk_rdev_t *rdev;
			rcu_read_lock();
			rdev = rcu_dereference(conf->disks[i].rdev);
			if (rdev && test_bit(In_sync, &rdev->flags))
2264 2265 2266
				atomic_inc(&rdev->nr_pending);
			else
				rdev = NULL;
2267
			rcu_read_unlock();
2268 2269 2270 2271 2272 2273 2274 2275
			if (rdev) {
				if (!rdev_set_badblocks(
					    rdev,
					    sh->sector,
					    STRIPE_SECTORS, 0))
					md_error(conf->mddev, rdev);
				rdev_dec_pending(rdev, conf->mddev);
			}
2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
		}
		spin_lock_irq(&conf->device_lock);
		/* fail all writes first */
		bi = sh->dev[i].towrite;
		sh->dev[i].towrite = NULL;
		if (bi) {
			s->to_write--;
			bitmap_end = 1;
		}

		if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
			wake_up(&conf->wait_for_overlap);

		while (bi && bi->bi_sector <
			sh->dev[i].sector + STRIPE_SECTORS) {
			struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
			clear_bit(BIO_UPTODATE, &bi->bi_flags);
2293
			if (!raid5_dec_bi_phys_segments(bi)) {
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307
				md_write_end(conf->mddev);
				bi->bi_next = *return_bi;
				*return_bi = bi;
			}
			bi = nextbi;
		}
		/* and fail all 'written' */
		bi = sh->dev[i].written;
		sh->dev[i].written = NULL;
		if (bi) bitmap_end = 1;
		while (bi && bi->bi_sector <
		       sh->dev[i].sector + STRIPE_SECTORS) {
			struct bio *bi2 = r5_next_bio(bi, sh->dev[i].sector);
			clear_bit(BIO_UPTODATE, &bi->bi_flags);
2308
			if (!raid5_dec_bi_phys_segments(bi)) {
2309 2310 2311 2312 2313 2314 2315
				md_write_end(conf->mddev);
				bi->bi_next = *return_bi;
				*return_bi = bi;
			}
			bi = bi2;
		}

2316 2317 2318 2319 2320 2321
		/* fail any reads if this device is non-operational and
		 * the data has not reached the cache yet.
		 */
		if (!test_bit(R5_Wantfill, &sh->dev[i].flags) &&
		    (!test_bit(R5_Insync, &sh->dev[i].flags) ||
		      test_bit(R5_ReadError, &sh->dev[i].flags))) {
2322 2323 2324 2325 2326 2327 2328 2329 2330 2331
			bi = sh->dev[i].toread;
			sh->dev[i].toread = NULL;
			if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
				wake_up(&conf->wait_for_overlap);
			if (bi) s->to_read--;
			while (bi && bi->bi_sector <
			       sh->dev[i].sector + STRIPE_SECTORS) {
				struct bio *nextbi =
					r5_next_bio(bi, sh->dev[i].sector);
				clear_bit(BIO_UPTODATE, &bi->bi_flags);
2332
				if (!raid5_dec_bi_phys_segments(bi)) {
2333 2334 2335 2336 2337 2338 2339 2340 2341 2342
					bi->bi_next = *return_bi;
					*return_bi = bi;
				}
				bi = nextbi;
			}
		}
		spin_unlock_irq(&conf->device_lock);
		if (bitmap_end)
			bitmap_endwrite(conf->mddev->bitmap, sh->sector,
					STRIPE_SECTORS, 0, 0);
2343 2344 2345 2346
		/* If we were in the middle of a write the parity block might
		 * still be locked - so just clear all R5_LOCKED flags
		 */
		clear_bit(R5_LOCKED, &sh->dev[i].flags);
2347 2348
	}

2349 2350 2351
	if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
		if (atomic_dec_and_test(&conf->pending_full_writes))
			md_wakeup_thread(conf->mddev->thread);
2352 2353
}

2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388
static void
handle_failed_sync(raid5_conf_t *conf, struct stripe_head *sh,
		   struct stripe_head_state *s)
{
	int abort = 0;
	int i;

	md_done_sync(conf->mddev, STRIPE_SECTORS, 0);
	clear_bit(STRIPE_SYNCING, &sh->state);
	s->syncing = 0;
	/* There is nothing more to do for sync/check/repair.
	 * For recover we need to record a bad block on all
	 * non-sync devices, or abort the recovery
	 */
	if (!test_bit(MD_RECOVERY_RECOVER, &conf->mddev->recovery))
		return;
	/* During recovery devices cannot be removed, so locking and
	 * refcounting of rdevs is not needed
	 */
	for (i = 0; i < conf->raid_disks; i++) {
		mdk_rdev_t *rdev = conf->disks[i].rdev;
		if (!rdev
		    || test_bit(Faulty, &rdev->flags)
		    || test_bit(In_sync, &rdev->flags))
			continue;
		if (!rdev_set_badblocks(rdev, sh->sector,
					STRIPE_SECTORS, 0))
			abort = 1;
	}
	if (abort) {
		conf->recovery_disabled = conf->mddev->recovery_disabled;
		set_bit(MD_RECOVERY_INTR, &conf->mddev->recovery);
	}
}

2389
/* fetch_block - checks the given member device to see if its data needs
2390 2391 2392
 * to be read or computed to satisfy a request.
 *
 * Returns 1 when no more member devices need to be checked, otherwise returns
2393
 * 0 to tell the loop in handle_stripe_fill to continue
2394
 */
2395 2396
static int fetch_block(struct stripe_head *sh, struct stripe_head_state *s,
		       int disk_idx, int disks)
2397
{
2398
	struct r5dev *dev = &sh->dev[disk_idx];
2399 2400
	struct r5dev *fdev[2] = { &sh->dev[s->failed_num[0]],
				  &sh->dev[s->failed_num[1]] };
2401

2402
	/* is the data in this block needed, and can we get it? */
2403 2404 2405 2406 2407
	if (!test_bit(R5_LOCKED, &dev->flags) &&
	    !test_bit(R5_UPTODATE, &dev->flags) &&
	    (dev->toread ||
	     (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags)) ||
	     s->syncing || s->expanding ||
2408 2409
	     (s->failed >= 1 && fdev[0]->toread) ||
	     (s->failed >= 2 && fdev[1]->toread) ||
2410 2411 2412
	     (sh->raid_conf->level <= 5 && s->failed && fdev[0]->towrite &&
	      !test_bit(R5_OVERWRITE, &fdev[0]->flags)) ||
	     (sh->raid_conf->level == 6 && s->failed && s->to_write))) {
2413 2414 2415 2416 2417 2418
		/* we would like to get this block, possibly by computing it,
		 * otherwise read it if the backing disk is insync
		 */
		BUG_ON(test_bit(R5_Wantcompute, &dev->flags));
		BUG_ON(test_bit(R5_Wantread, &dev->flags));
		if ((s->uptodate == disks - 1) &&
2419 2420
		    (s->failed && (disk_idx == s->failed_num[0] ||
				   disk_idx == s->failed_num[1]))) {
2421 2422
			/* have disk failed, and we're requested to fetch it;
			 * do compute it
2423
			 */
2424 2425 2426 2427 2428 2429 2430 2431
			pr_debug("Computing stripe %llu block %d\n",
			       (unsigned long long)sh->sector, disk_idx);
			set_bit(STRIPE_COMPUTE_RUN, &sh->state);
			set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
			set_bit(R5_Wantcompute, &dev->flags);
			sh->ops.target = disk_idx;
			sh->ops.target2 = -1; /* no 2nd target */
			s->req_compute = 1;
2432 2433 2434 2435 2436 2437
			/* Careful: from this point on 'uptodate' is in the eye
			 * of raid_run_ops which services 'compute' operations
			 * before writes. R5_Wantcompute flags a block that will
			 * be R5_UPTODATE by the time it is needed for a
			 * subsequent operation.
			 */
2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
			s->uptodate++;
			return 1;
		} else if (s->uptodate == disks-2 && s->failed >= 2) {
			/* Computing 2-failure is *very* expensive; only
			 * do it if failed >= 2
			 */
			int other;
			for (other = disks; other--; ) {
				if (other == disk_idx)
					continue;
				if (!test_bit(R5_UPTODATE,
				      &sh->dev[other].flags))
					break;
2451
			}
2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470
			BUG_ON(other < 0);
			pr_debug("Computing stripe %llu blocks %d,%d\n",
			       (unsigned long long)sh->sector,
			       disk_idx, other);
			set_bit(STRIPE_COMPUTE_RUN, &sh->state);
			set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
			set_bit(R5_Wantcompute, &sh->dev[disk_idx].flags);
			set_bit(R5_Wantcompute, &sh->dev[other].flags);
			sh->ops.target = disk_idx;
			sh->ops.target2 = other;
			s->uptodate += 2;
			s->req_compute = 1;
			return 1;
		} else if (test_bit(R5_Insync, &dev->flags)) {
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantread, &dev->flags);
			s->locked++;
			pr_debug("Reading block %d (sync=%d)\n",
				disk_idx, s->syncing);
2471 2472
		}
	}
2473 2474 2475 2476 2477

	return 0;
}

/**
2478
 * handle_stripe_fill - read or compute data to satisfy pending requests.
2479
 */
2480 2481 2482
static void handle_stripe_fill(struct stripe_head *sh,
			       struct stripe_head_state *s,
			       int disks)
2483 2484 2485 2486 2487 2488 2489 2490 2491 2492
{
	int i;

	/* look for blocks to read/compute, skip this if a compute
	 * is already in flight, or if the stripe contents are in the
	 * midst of changing due to a write
	 */
	if (!test_bit(STRIPE_COMPUTE_RUN, &sh->state) && !sh->check_state &&
	    !sh->reconstruct_state)
		for (i = disks; i--; )
2493
			if (fetch_block(sh, s, i, disks))
2494
				break;
2495 2496 2497 2498
	set_bit(STRIPE_HANDLE, &sh->state);
}


2499
/* handle_stripe_clean_event
2500 2501 2502 2503
 * any written block on an uptodate or failed drive can be returned.
 * Note that if we 'wrote' to a failed drive, it will be UPTODATE, but
 * never LOCKED, so we don't need to test 'failed' directly.
 */
2504
static void handle_stripe_clean_event(raid5_conf_t *conf,
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517
	struct stripe_head *sh, int disks, struct bio **return_bi)
{
	int i;
	struct r5dev *dev;

	for (i = disks; i--; )
		if (sh->dev[i].written) {
			dev = &sh->dev[i];
			if (!test_bit(R5_LOCKED, &dev->flags) &&
				test_bit(R5_UPTODATE, &dev->flags)) {
				/* We can return any write requests */
				struct bio *wbi, *wbi2;
				int bitmap_end = 0;
2518
				pr_debug("Return write for disc %d\n", i);
2519 2520 2521 2522 2523 2524
				spin_lock_irq(&conf->device_lock);
				wbi = dev->written;
				dev->written = NULL;
				while (wbi && wbi->bi_sector <
					dev->sector + STRIPE_SECTORS) {
					wbi2 = r5_next_bio(wbi, dev->sector);
2525
					if (!raid5_dec_bi_phys_segments(wbi)) {
2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542
						md_write_end(conf->mddev);
						wbi->bi_next = *return_bi;
						*return_bi = wbi;
					}
					wbi = wbi2;
				}
				if (dev->towrite == NULL)
					bitmap_end = 1;
				spin_unlock_irq(&conf->device_lock);
				if (bitmap_end)
					bitmap_endwrite(conf->mddev->bitmap,
							sh->sector,
							STRIPE_SECTORS,
					 !test_bit(STRIPE_DEGRADED, &sh->state),
							0);
			}
		}
2543 2544 2545 2546

	if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
		if (atomic_dec_and_test(&conf->pending_full_writes))
			md_wakeup_thread(conf->mddev->thread);
2547 2548
}

2549 2550 2551 2552
static void handle_stripe_dirtying(raid5_conf_t *conf,
				   struct stripe_head *sh,
				   struct stripe_head_state *s,
				   int disks)
2553 2554
{
	int rmw = 0, rcw = 0, i;
2555 2556 2557 2558 2559 2560 2561
	if (conf->max_degraded == 2) {
		/* RAID6 requires 'rcw' in current implementation
		 * Calculate the real rcw later - for now fake it
		 * look like rcw is cheaper
		 */
		rcw = 1; rmw = 2;
	} else for (i = disks; i--; ) {
2562 2563 2564 2565
		/* would I have to read this buffer for read_modify_write */
		struct r5dev *dev = &sh->dev[i];
		if ((dev->towrite || i == sh->pd_idx) &&
		    !test_bit(R5_LOCKED, &dev->flags) &&
2566 2567
		    !(test_bit(R5_UPTODATE, &dev->flags) ||
		      test_bit(R5_Wantcompute, &dev->flags))) {
2568 2569 2570 2571 2572 2573 2574 2575
			if (test_bit(R5_Insync, &dev->flags))
				rmw++;
			else
				rmw += 2*disks;  /* cannot read it */
		}
		/* Would I have to read this buffer for reconstruct_write */
		if (!test_bit(R5_OVERWRITE, &dev->flags) && i != sh->pd_idx &&
		    !test_bit(R5_LOCKED, &dev->flags) &&
2576 2577 2578
		    !(test_bit(R5_UPTODATE, &dev->flags) ||
		    test_bit(R5_Wantcompute, &dev->flags))) {
			if (test_bit(R5_Insync, &dev->flags)) rcw++;
2579 2580 2581 2582
			else
				rcw += 2*disks;
		}
	}
2583
	pr_debug("for sector %llu, rmw=%d rcw=%d\n",
2584 2585 2586 2587 2588 2589 2590 2591
		(unsigned long long)sh->sector, rmw, rcw);
	set_bit(STRIPE_HANDLE, &sh->state);
	if (rmw < rcw && rmw > 0)
		/* prefer read-modify-write, but need to get some data */
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if ((dev->towrite || i == sh->pd_idx) &&
			    !test_bit(R5_LOCKED, &dev->flags) &&
2592 2593
			    !(test_bit(R5_UPTODATE, &dev->flags) ||
			    test_bit(R5_Wantcompute, &dev->flags)) &&
2594 2595 2596
			    test_bit(R5_Insync, &dev->flags)) {
				if (
				  test_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
2597
					pr_debug("Read_old block "
2598 2599 2600 2601 2602 2603 2604 2605 2606 2607
						"%d for r-m-w\n", i);
					set_bit(R5_LOCKED, &dev->flags);
					set_bit(R5_Wantread, &dev->flags);
					s->locked++;
				} else {
					set_bit(STRIPE_DELAYED, &sh->state);
					set_bit(STRIPE_HANDLE, &sh->state);
				}
			}
		}
2608
	if (rcw <= rmw && rcw > 0) {
2609
		/* want reconstruct write, but need to get some data */
2610
		rcw = 0;
2611 2612 2613
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (!test_bit(R5_OVERWRITE, &dev->flags) &&
2614
			    i != sh->pd_idx && i != sh->qd_idx &&
2615
			    !test_bit(R5_LOCKED, &dev->flags) &&
2616
			    !(test_bit(R5_UPTODATE, &dev->flags) ||
2617 2618 2619 2620
			      test_bit(R5_Wantcompute, &dev->flags))) {
				rcw++;
				if (!test_bit(R5_Insync, &dev->flags))
					continue; /* it's a failed drive */
2621 2622
				if (
				  test_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
2623
					pr_debug("Read_old block "
2624 2625 2626 2627 2628 2629 2630 2631 2632 2633
						"%d for Reconstruct\n", i);
					set_bit(R5_LOCKED, &dev->flags);
					set_bit(R5_Wantread, &dev->flags);
					s->locked++;
				} else {
					set_bit(STRIPE_DELAYED, &sh->state);
					set_bit(STRIPE_HANDLE, &sh->state);
				}
			}
		}
2634
	}
2635 2636 2637
	/* now if nothing is locked, and if we have enough data,
	 * we can start a write request
	 */
2638 2639
	/* since handle_stripe can be called at any time we need to handle the
	 * case where a compute block operation has been submitted and then a
2640 2641
	 * subsequent call wants to start a write request.  raid_run_ops only
	 * handles the case where compute block and reconstruct are requested
2642 2643 2644
	 * simultaneously.  If this is not the case then new writes need to be
	 * held off until the compute completes.
	 */
2645 2646 2647
	if ((s->req_compute || !test_bit(STRIPE_COMPUTE_RUN, &sh->state)) &&
	    (s->locked == 0 && (rcw == 0 || rmw == 0) &&
	    !test_bit(STRIPE_BIT_DELAY, &sh->state)))
2648
		schedule_reconstruction(sh, s, rcw == 0, 0);
2649 2650 2651 2652 2653
}

static void handle_parity_checks5(raid5_conf_t *conf, struct stripe_head *sh,
				struct stripe_head_state *s, int disks)
{
2654
	struct r5dev *dev = NULL;
2655

2656
	set_bit(STRIPE_HANDLE, &sh->state);
2657

2658 2659 2660
	switch (sh->check_state) {
	case check_state_idle:
		/* start a new check operation if there are no failures */
2661 2662
		if (s->failed == 0) {
			BUG_ON(s->uptodate != disks);
2663 2664
			sh->check_state = check_state_run;
			set_bit(STRIPE_OP_CHECK, &s->ops_request);
2665 2666
			clear_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags);
			s->uptodate--;
2667
			break;
2668
		}
2669
		dev = &sh->dev[s->failed_num[0]];
2670 2671 2672 2673 2674 2675 2676 2677 2678
		/* fall through */
	case check_state_compute_result:
		sh->check_state = check_state_idle;
		if (!dev)
			dev = &sh->dev[sh->pd_idx];

		/* check that a write has not made the stripe insync */
		if (test_bit(STRIPE_INSYNC, &sh->state))
			break;
D
Dan Williams 已提交
2679

2680 2681 2682 2683 2684
		/* either failed parity check, or recovery is happening */
		BUG_ON(!test_bit(R5_UPTODATE, &dev->flags));
		BUG_ON(s->uptodate != disks);

		set_bit(R5_LOCKED, &dev->flags);
2685
		s->locked++;
2686
		set_bit(R5_Wantwrite, &dev->flags);
2687

2688 2689
		clear_bit(STRIPE_DEGRADED, &sh->state);
		set_bit(STRIPE_INSYNC, &sh->state);
2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705
		break;
	case check_state_run:
		break; /* we will be called again upon completion */
	case check_state_check_result:
		sh->check_state = check_state_idle;

		/* if a failure occurred during the check operation, leave
		 * STRIPE_INSYNC not set and let the stripe be handled again
		 */
		if (s->failed)
			break;

		/* handle a successful check operation, if parity is correct
		 * we are done.  Otherwise update the mismatch count and repair
		 * parity if !MD_RECOVERY_CHECK
		 */
D
Dan Williams 已提交
2706
		if ((sh->ops.zero_sum_result & SUM_CHECK_P_RESULT) == 0)
2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717
			/* parity is correct (on disc,
			 * not in buffer any more)
			 */
			set_bit(STRIPE_INSYNC, &sh->state);
		else {
			conf->mddev->resync_mismatches += STRIPE_SECTORS;
			if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
				/* don't try to repair!! */
				set_bit(STRIPE_INSYNC, &sh->state);
			else {
				sh->check_state = check_state_compute_run;
2718
				set_bit(STRIPE_COMPUTE_RUN, &sh->state);
2719 2720 2721 2722
				set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
				set_bit(R5_Wantcompute,
					&sh->dev[sh->pd_idx].flags);
				sh->ops.target = sh->pd_idx;
2723
				sh->ops.target2 = -1;
2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734
				s->uptodate++;
			}
		}
		break;
	case check_state_compute_run:
		break;
	default:
		printk(KERN_ERR "%s: unknown check_state: %d sector: %llu\n",
		       __func__, sh->check_state,
		       (unsigned long long) sh->sector);
		BUG();
2735 2736 2737 2738 2739
	}
}


static void handle_parity_checks6(raid5_conf_t *conf, struct stripe_head *sh,
2740
				  struct stripe_head_state *s,
2741
				  int disks)
2742 2743
{
	int pd_idx = sh->pd_idx;
N
NeilBrown 已提交
2744
	int qd_idx = sh->qd_idx;
2745
	struct r5dev *dev;
2746 2747 2748 2749

	set_bit(STRIPE_HANDLE, &sh->state);

	BUG_ON(s->failed > 2);
2750

2751 2752 2753 2754 2755 2756
	/* Want to check and possibly repair P and Q.
	 * However there could be one 'failed' device, in which
	 * case we can only check one of them, possibly using the
	 * other to generate missing data
	 */

2757 2758 2759
	switch (sh->check_state) {
	case check_state_idle:
		/* start a new check operation if there are < 2 failures */
2760
		if (s->failed == s->q_failed) {
2761
			/* The only possible failed device holds Q, so it
2762 2763 2764
			 * makes sense to check P (If anything else were failed,
			 * we would have used P to recreate it).
			 */
2765
			sh->check_state = check_state_run;
2766
		}
2767
		if (!s->q_failed && s->failed < 2) {
2768
			/* Q is not failed, and we didn't use it to generate
2769 2770
			 * anything, so it makes sense to check it
			 */
2771 2772 2773 2774
			if (sh->check_state == check_state_run)
				sh->check_state = check_state_run_pq;
			else
				sh->check_state = check_state_run_q;
2775 2776
		}

2777 2778
		/* discard potentially stale zero_sum_result */
		sh->ops.zero_sum_result = 0;
2779

2780 2781 2782 2783
		if (sh->check_state == check_state_run) {
			/* async_xor_zero_sum destroys the contents of P */
			clear_bit(R5_UPTODATE, &sh->dev[pd_idx].flags);
			s->uptodate--;
2784
		}
2785 2786 2787 2788 2789 2790 2791
		if (sh->check_state >= check_state_run &&
		    sh->check_state <= check_state_run_pq) {
			/* async_syndrome_zero_sum preserves P and Q, so
			 * no need to mark them !uptodate here
			 */
			set_bit(STRIPE_OP_CHECK, &s->ops_request);
			break;
2792 2793
		}

2794 2795 2796 2797 2798
		/* we have 2-disk failure */
		BUG_ON(s->failed != 2);
		/* fall through */
	case check_state_compute_result:
		sh->check_state = check_state_idle;
2799

2800 2801 2802
		/* check that a write has not made the stripe insync */
		if (test_bit(STRIPE_INSYNC, &sh->state))
			break;
2803 2804

		/* now write out any block on a failed drive,
2805
		 * or P or Q if they were recomputed
2806
		 */
2807
		BUG_ON(s->uptodate < disks - 1); /* We don't need Q to recover */
2808
		if (s->failed == 2) {
2809
			dev = &sh->dev[s->failed_num[1]];
2810 2811 2812 2813 2814
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
		if (s->failed >= 1) {
2815
			dev = &sh->dev[s->failed_num[0]];
2816 2817 2818 2819
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
2820
		if (sh->ops.zero_sum_result & SUM_CHECK_P_RESULT) {
2821 2822 2823 2824 2825
			dev = &sh->dev[pd_idx];
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
2826
		if (sh->ops.zero_sum_result & SUM_CHECK_Q_RESULT) {
2827 2828 2829 2830 2831 2832 2833 2834
			dev = &sh->dev[qd_idx];
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
		clear_bit(STRIPE_DEGRADED, &sh->state);

		set_bit(STRIPE_INSYNC, &sh->state);
2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898
		break;
	case check_state_run:
	case check_state_run_q:
	case check_state_run_pq:
		break; /* we will be called again upon completion */
	case check_state_check_result:
		sh->check_state = check_state_idle;

		/* handle a successful check operation, if parity is correct
		 * we are done.  Otherwise update the mismatch count and repair
		 * parity if !MD_RECOVERY_CHECK
		 */
		if (sh->ops.zero_sum_result == 0) {
			/* both parities are correct */
			if (!s->failed)
				set_bit(STRIPE_INSYNC, &sh->state);
			else {
				/* in contrast to the raid5 case we can validate
				 * parity, but still have a failure to write
				 * back
				 */
				sh->check_state = check_state_compute_result;
				/* Returning at this point means that we may go
				 * off and bring p and/or q uptodate again so
				 * we make sure to check zero_sum_result again
				 * to verify if p or q need writeback
				 */
			}
		} else {
			conf->mddev->resync_mismatches += STRIPE_SECTORS;
			if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
				/* don't try to repair!! */
				set_bit(STRIPE_INSYNC, &sh->state);
			else {
				int *target = &sh->ops.target;

				sh->ops.target = -1;
				sh->ops.target2 = -1;
				sh->check_state = check_state_compute_run;
				set_bit(STRIPE_COMPUTE_RUN, &sh->state);
				set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
				if (sh->ops.zero_sum_result & SUM_CHECK_P_RESULT) {
					set_bit(R5_Wantcompute,
						&sh->dev[pd_idx].flags);
					*target = pd_idx;
					target = &sh->ops.target2;
					s->uptodate++;
				}
				if (sh->ops.zero_sum_result & SUM_CHECK_Q_RESULT) {
					set_bit(R5_Wantcompute,
						&sh->dev[qd_idx].flags);
					*target = qd_idx;
					s->uptodate++;
				}
			}
		}
		break;
	case check_state_compute_run:
		break;
	default:
		printk(KERN_ERR "%s: unknown check_state: %d sector: %llu\n",
		       __func__, sh->check_state,
		       (unsigned long long) sh->sector);
		BUG();
2899 2900 2901
	}
}

2902
static void handle_stripe_expansion(raid5_conf_t *conf, struct stripe_head *sh)
2903 2904 2905 2906 2907 2908
{
	int i;

	/* We have read all the blocks in this stripe and now we need to
	 * copy some of them into a target stripe for expand.
	 */
2909
	struct dma_async_tx_descriptor *tx = NULL;
2910 2911
	clear_bit(STRIPE_EXPAND_SOURCE, &sh->state);
	for (i = 0; i < sh->disks; i++)
N
NeilBrown 已提交
2912
		if (i != sh->pd_idx && i != sh->qd_idx) {
2913
			int dd_idx, j;
2914
			struct stripe_head *sh2;
2915
			struct async_submit_ctl submit;
2916

2917
			sector_t bn = compute_blocknr(sh, i, 1);
2918 2919
			sector_t s = raid5_compute_sector(conf, bn, 0,
							  &dd_idx, NULL);
2920
			sh2 = get_active_stripe(conf, s, 0, 1, 1);
2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932
			if (sh2 == NULL)
				/* so far only the early blocks of this stripe
				 * have been requested.  When later blocks
				 * get requested, we will try again
				 */
				continue;
			if (!test_bit(STRIPE_EXPANDING, &sh2->state) ||
			   test_bit(R5_Expanded, &sh2->dev[dd_idx].flags)) {
				/* must have already done this block */
				release_stripe(sh2);
				continue;
			}
2933 2934

			/* place all the copies on one channel */
2935
			init_async_submit(&submit, 0, tx, NULL, NULL, NULL);
2936
			tx = async_memcpy(sh2->dev[dd_idx].page,
2937
					  sh->dev[i].page, 0, 0, STRIPE_SIZE,
2938
					  &submit);
2939

2940 2941 2942 2943
			set_bit(R5_Expanded, &sh2->dev[dd_idx].flags);
			set_bit(R5_UPTODATE, &sh2->dev[dd_idx].flags);
			for (j = 0; j < conf->raid_disks; j++)
				if (j != sh2->pd_idx &&
2944
				    j != sh2->qd_idx &&
2945 2946 2947 2948 2949 2950 2951
				    !test_bit(R5_Expanded, &sh2->dev[j].flags))
					break;
			if (j == conf->raid_disks) {
				set_bit(STRIPE_EXPAND_READY, &sh2->state);
				set_bit(STRIPE_HANDLE, &sh2->state);
			}
			release_stripe(sh2);
2952

2953
		}
2954 2955 2956 2957 2958
	/* done submitting copies, wait for them to complete */
	if (tx) {
		async_tx_ack(tx);
		dma_wait_for_async_tx(tx);
	}
2959
}
L
Linus Torvalds 已提交
2960

2961

L
Linus Torvalds 已提交
2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977
/*
 * handle_stripe - do things to a stripe.
 *
 * We lock the stripe and then examine the state of various bits
 * to see what needs to be done.
 * Possible results:
 *    return some read request which now have data
 *    return some write requests which are safely on disc
 *    schedule a read on some buffers
 *    schedule a write of some buffers
 *    return confirmation of parity correctness
 *
 * buffers are taken off read_list or write_list, and bh_cache buffers
 * get BH_Lock set before the stripe lock is released.
 *
 */
2978

2979
static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s)
L
Linus Torvalds 已提交
2980
{
2981
	raid5_conf_t *conf = sh->raid_conf;
2982
	int disks = sh->disks;
2983 2984
	struct r5dev *dev;
	int i;
L
Linus Torvalds 已提交
2985

2986 2987 2988 2989 2990 2991 2992
	memset(s, 0, sizeof(*s));

	s->syncing = test_bit(STRIPE_SYNCING, &sh->state);
	s->expanding = test_bit(STRIPE_EXPAND_SOURCE, &sh->state);
	s->expanded = test_bit(STRIPE_EXPAND_READY, &sh->state);
	s->failed_num[0] = -1;
	s->failed_num[1] = -1;
L
Linus Torvalds 已提交
2993

2994
	/* Now to look around and see what can be done */
L
Linus Torvalds 已提交
2995
	rcu_read_lock();
2996
	spin_lock_irq(&conf->device_lock);
2997 2998
	for (i=disks; i--; ) {
		mdk_rdev_t *rdev;
2999 3000 3001
		sector_t first_bad;
		int bad_sectors;
		int is_bad = 0;
3002

3003
		dev = &sh->dev[i];
L
Linus Torvalds 已提交
3004

3005
		pr_debug("check %d: state 0x%lx read %p write %p written %p\n",
3006
			i, dev->flags, dev->toread, dev->towrite, dev->written);
3007 3008 3009 3010 3011 3012 3013 3014
		/* maybe we can reply to a read
		 *
		 * new wantfill requests are only permitted while
		 * ops_complete_biofill is guaranteed to be inactive
		 */
		if (test_bit(R5_UPTODATE, &dev->flags) && dev->toread &&
		    !test_bit(STRIPE_BIOFILL_RUN, &sh->state))
			set_bit(R5_Wantfill, &dev->flags);
L
Linus Torvalds 已提交
3015

3016
		/* now count some things */
3017 3018 3019 3020
		if (test_bit(R5_LOCKED, &dev->flags))
			s->locked++;
		if (test_bit(R5_UPTODATE, &dev->flags))
			s->uptodate++;
3021
		if (test_bit(R5_Wantcompute, &dev->flags)) {
3022 3023
			s->compute++;
			BUG_ON(s->compute > 2);
3024
		}
L
Linus Torvalds 已提交
3025

3026
		if (test_bit(R5_Wantfill, &dev->flags))
3027
			s->to_fill++;
3028
		else if (dev->toread)
3029
			s->to_read++;
3030
		if (dev->towrite) {
3031
			s->to_write++;
3032
			if (!test_bit(R5_OVERWRITE, &dev->flags))
3033
				s->non_overwrite++;
3034
		}
3035
		if (dev->written)
3036
			s->written++;
3037
		rdev = rcu_dereference(conf->disks[i].rdev);
3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049
		if (rdev) {
			is_bad = is_badblock(rdev, sh->sector, STRIPE_SECTORS,
					     &first_bad, &bad_sectors);
			if (s->blocked_rdev == NULL
			    && (test_bit(Blocked, &rdev->flags)
				|| is_bad < 0)) {
				if (is_bad < 0)
					set_bit(BlockedBadBlocks,
						&rdev->flags);
				s->blocked_rdev = rdev;
				atomic_inc(&rdev->nr_pending);
			}
3050
		}
3051 3052 3053
		clear_bit(R5_Insync, &dev->flags);
		if (!rdev)
			/* Not in-sync */;
3054 3055 3056 3057 3058 3059 3060 3061 3062 3063
		else if (is_bad) {
			/* also not in-sync */
			if (!test_bit(WriteErrorSeen, &rdev->flags)) {
				/* treat as in-sync, but with a read error
				 * which we can now try to correct
				 */
				set_bit(R5_Insync, &dev->flags);
				set_bit(R5_ReadError, &dev->flags);
			}
		} else if (test_bit(In_sync, &rdev->flags))
3064 3065 3066 3067 3068 3069
			set_bit(R5_Insync, &dev->flags);
		else {
			/* in sync if before recovery_offset */
			if (sh->sector + STRIPE_SECTORS <= rdev->recovery_offset)
				set_bit(R5_Insync, &dev->flags);
		}
3070 3071 3072 3073 3074 3075 3076 3077
		if (test_bit(R5_WriteError, &dev->flags)) {
			clear_bit(R5_Insync, &dev->flags);
			if (!test_bit(Faulty, &rdev->flags)) {
				s->handle_bad_blocks = 1;
				atomic_inc(&rdev->nr_pending);
			} else
				clear_bit(R5_WriteError, &dev->flags);
		}
3078 3079 3080 3081 3082 3083 3084
		if (test_bit(R5_MadeGood, &dev->flags)) {
			if (!test_bit(Faulty, &rdev->flags)) {
				s->handle_bad_blocks = 1;
				atomic_inc(&rdev->nr_pending);
			} else
				clear_bit(R5_MadeGood, &dev->flags);
		}
3085
		if (!test_bit(R5_Insync, &dev->flags)) {
3086 3087 3088
			/* The ReadError flag will just be confusing now */
			clear_bit(R5_ReadError, &dev->flags);
			clear_bit(R5_ReWrite, &dev->flags);
L
Linus Torvalds 已提交
3089
		}
3090 3091 3092
		if (test_bit(R5_ReadError, &dev->flags))
			clear_bit(R5_Insync, &dev->flags);
		if (!test_bit(R5_Insync, &dev->flags)) {
3093 3094 3095
			if (s->failed < 2)
				s->failed_num[s->failed] = i;
			s->failed++;
3096
		}
L
Linus Torvalds 已提交
3097
	}
3098
	spin_unlock_irq(&conf->device_lock);
L
Linus Torvalds 已提交
3099
	rcu_read_unlock();
3100 3101 3102 3103 3104
}

static void handle_stripe(struct stripe_head *sh)
{
	struct stripe_head_state s;
3105
	raid5_conf_t *conf = sh->raid_conf;
3106
	int i;
3107 3108
	int prexor;
	int disks = sh->disks;
3109
	struct r5dev *pdev, *qdev;
3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129

	clear_bit(STRIPE_HANDLE, &sh->state);
	if (test_and_set_bit(STRIPE_ACTIVE, &sh->state)) {
		/* already being handled, ensure it gets handled
		 * again when current action finishes */
		set_bit(STRIPE_HANDLE, &sh->state);
		return;
	}

	if (test_and_clear_bit(STRIPE_SYNC_REQUESTED, &sh->state)) {
		set_bit(STRIPE_SYNCING, &sh->state);
		clear_bit(STRIPE_INSYNC, &sh->state);
	}
	clear_bit(STRIPE_DELAYED, &sh->state);

	pr_debug("handling stripe %llu, state=%#lx cnt=%d, "
		"pd_idx=%d, qd_idx=%d\n, check:%d, reconstruct:%d\n",
	       (unsigned long long)sh->sector, sh->state,
	       atomic_read(&sh->count), sh->pd_idx, sh->qd_idx,
	       sh->check_state, sh->reconstruct_state);
3130

3131
	analyse_stripe(sh, &s);
3132

3133 3134 3135 3136 3137
	if (s.handle_bad_blocks) {
		set_bit(STRIPE_HANDLE, &sh->state);
		goto finish;
	}

3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162
	if (unlikely(s.blocked_rdev)) {
		if (s.syncing || s.expanding || s.expanded ||
		    s.to_write || s.written) {
			set_bit(STRIPE_HANDLE, &sh->state);
			goto finish;
		}
		/* There is nothing for the blocked_rdev to block */
		rdev_dec_pending(s.blocked_rdev, conf->mddev);
		s.blocked_rdev = NULL;
	}

	if (s.to_fill && !test_bit(STRIPE_BIOFILL_RUN, &sh->state)) {
		set_bit(STRIPE_OP_BIOFILL, &s.ops_request);
		set_bit(STRIPE_BIOFILL_RUN, &sh->state);
	}

	pr_debug("locked=%d uptodate=%d to_read=%d"
	       " to_write=%d failed=%d failed_num=%d,%d\n",
	       s.locked, s.uptodate, s.to_read, s.to_write, s.failed,
	       s.failed_num[0], s.failed_num[1]);
	/* check if the array has lost more than max_degraded devices and,
	 * if so, some requests might need to be failed.
	 */
	if (s.failed > conf->max_degraded && s.to_read+s.to_write+s.written)
		handle_failed_stripe(conf, sh, &s, disks, &s.return_bi);
3163 3164
	if (s.failed > conf->max_degraded && s.syncing)
		handle_failed_sync(conf, sh, &s);
3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195

	/*
	 * might be able to return some write requests if the parity blocks
	 * are safe, or on a failed drive
	 */
	pdev = &sh->dev[sh->pd_idx];
	s.p_failed = (s.failed >= 1 && s.failed_num[0] == sh->pd_idx)
		|| (s.failed >= 2 && s.failed_num[1] == sh->pd_idx);
	qdev = &sh->dev[sh->qd_idx];
	s.q_failed = (s.failed >= 1 && s.failed_num[0] == sh->qd_idx)
		|| (s.failed >= 2 && s.failed_num[1] == sh->qd_idx)
		|| conf->level < 6;

	if (s.written &&
	    (s.p_failed || ((test_bit(R5_Insync, &pdev->flags)
			     && !test_bit(R5_LOCKED, &pdev->flags)
			     && test_bit(R5_UPTODATE, &pdev->flags)))) &&
	    (s.q_failed || ((test_bit(R5_Insync, &qdev->flags)
			     && !test_bit(R5_LOCKED, &qdev->flags)
			     && test_bit(R5_UPTODATE, &qdev->flags)))))
		handle_stripe_clean_event(conf, sh, disks, &s.return_bi);

	/* Now we might consider reading some blocks, either to check/generate
	 * parity, or to satisfy requests
	 * or to load a block that is being partially written.
	 */
	if (s.to_read || s.non_overwrite
	    || (conf->level == 6 && s.to_write && s.failed)
	    || (s.syncing && (s.uptodate + s.compute < disks)) || s.expanding)
		handle_stripe_fill(sh, &s, disks);

3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253
	/* Now we check to see if any write operations have recently
	 * completed
	 */
	prexor = 0;
	if (sh->reconstruct_state == reconstruct_state_prexor_drain_result)
		prexor = 1;
	if (sh->reconstruct_state == reconstruct_state_drain_result ||
	    sh->reconstruct_state == reconstruct_state_prexor_drain_result) {
		sh->reconstruct_state = reconstruct_state_idle;

		/* All the 'written' buffers and the parity block are ready to
		 * be written back to disk
		 */
		BUG_ON(!test_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags));
		BUG_ON(sh->qd_idx >= 0 &&
		       !test_bit(R5_UPTODATE, &sh->dev[sh->qd_idx].flags));
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (test_bit(R5_LOCKED, &dev->flags) &&
				(i == sh->pd_idx || i == sh->qd_idx ||
				 dev->written)) {
				pr_debug("Writing block %d\n", i);
				set_bit(R5_Wantwrite, &dev->flags);
				if (prexor)
					continue;
				if (!test_bit(R5_Insync, &dev->flags) ||
				    ((i == sh->pd_idx || i == sh->qd_idx)  &&
				     s.failed == 0))
					set_bit(STRIPE_INSYNC, &sh->state);
			}
		}
		if (test_and_clear_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
			s.dec_preread_active = 1;
	}

	/* Now to consider new write requests and what else, if anything
	 * should be read.  We do not handle new writes when:
	 * 1/ A 'write' operation (copy+xor) is already in flight.
	 * 2/ A 'check' operation is in flight, as it may clobber the parity
	 *    block.
	 */
	if (s.to_write && !sh->reconstruct_state && !sh->check_state)
		handle_stripe_dirtying(conf, sh, &s, disks);

	/* maybe we need to check and possibly fix the parity for this stripe
	 * Any reads will already have been scheduled, so we just see if enough
	 * data is available.  The parity check is held off while parity
	 * dependent operations are in flight.
	 */
	if (sh->check_state ||
	    (s.syncing && s.locked == 0 &&
	     !test_bit(STRIPE_COMPUTE_RUN, &sh->state) &&
	     !test_bit(STRIPE_INSYNC, &sh->state))) {
		if (conf->level == 6)
			handle_parity_checks6(conf, sh, &s, disks);
		else
			handle_parity_checks5(conf, sh, &s, disks);
	}
3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284

	if (s.syncing && s.locked == 0 && test_bit(STRIPE_INSYNC, &sh->state)) {
		md_done_sync(conf->mddev, STRIPE_SECTORS, 1);
		clear_bit(STRIPE_SYNCING, &sh->state);
	}

	/* If the failed drives are just a ReadError, then we might need
	 * to progress the repair/check process
	 */
	if (s.failed <= conf->max_degraded && !conf->mddev->ro)
		for (i = 0; i < s.failed; i++) {
			struct r5dev *dev = &sh->dev[s.failed_num[i]];
			if (test_bit(R5_ReadError, &dev->flags)
			    && !test_bit(R5_LOCKED, &dev->flags)
			    && test_bit(R5_UPTODATE, &dev->flags)
				) {
				if (!test_bit(R5_ReWrite, &dev->flags)) {
					set_bit(R5_Wantwrite, &dev->flags);
					set_bit(R5_ReWrite, &dev->flags);
					set_bit(R5_LOCKED, &dev->flags);
					s.locked++;
				} else {
					/* let's read it back */
					set_bit(R5_Wantread, &dev->flags);
					set_bit(R5_LOCKED, &dev->flags);
					s.locked++;
				}
			}
		}


3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311
	/* Finish reconstruct operations initiated by the expansion process */
	if (sh->reconstruct_state == reconstruct_state_result) {
		struct stripe_head *sh_src
			= get_active_stripe(conf, sh->sector, 1, 1, 1);
		if (sh_src && test_bit(STRIPE_EXPAND_SOURCE, &sh_src->state)) {
			/* sh cannot be written until sh_src has been read.
			 * so arrange for sh to be delayed a little
			 */
			set_bit(STRIPE_DELAYED, &sh->state);
			set_bit(STRIPE_HANDLE, &sh->state);
			if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE,
					      &sh_src->state))
				atomic_inc(&conf->preread_active_stripes);
			release_stripe(sh_src);
			goto finish;
		}
		if (sh_src)
			release_stripe(sh_src);

		sh->reconstruct_state = reconstruct_state_idle;
		clear_bit(STRIPE_EXPANDING, &sh->state);
		for (i = conf->raid_disks; i--; ) {
			set_bit(R5_Wantwrite, &sh->dev[i].flags);
			set_bit(R5_LOCKED, &sh->dev[i].flags);
			s.locked++;
		}
	}
3312

3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328
	if (s.expanded && test_bit(STRIPE_EXPANDING, &sh->state) &&
	    !sh->reconstruct_state) {
		/* Need to write out all blocks after computing parity */
		sh->disks = conf->raid_disks;
		stripe_set_idx(sh->sector, conf, 0, sh);
		schedule_reconstruction(sh, &s, 1, 1);
	} else if (s.expanded && !sh->reconstruct_state && s.locked == 0) {
		clear_bit(STRIPE_EXPAND_READY, &sh->state);
		atomic_dec(&conf->reshape_stripes);
		wake_up(&conf->wait_for_overlap);
		md_done_sync(conf->mddev, STRIPE_SECTORS, 1);
	}

	if (s.expanding && s.locked == 0 &&
	    !test_bit(STRIPE_COMPUTE_RUN, &sh->state))
		handle_stripe_expansion(conf, sh);
3329

3330
finish:
3331
	/* wait for this device to become unblocked */
3332
	if (conf->mddev->external && unlikely(s.blocked_rdev))
3333
		md_wait_for_blocked_rdev(s.blocked_rdev, conf->mddev);
3334

3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346
	if (s.handle_bad_blocks)
		for (i = disks; i--; ) {
			mdk_rdev_t *rdev;
			struct r5dev *dev = &sh->dev[i];
			if (test_and_clear_bit(R5_WriteError, &dev->flags)) {
				/* We own a safe reference to the rdev */
				rdev = conf->disks[i].rdev;
				if (!rdev_set_badblocks(rdev, sh->sector,
							STRIPE_SECTORS, 0))
					md_error(conf->mddev, rdev);
				rdev_dec_pending(rdev, conf->mddev);
			}
3347 3348 3349 3350 3351 3352
			if (test_and_clear_bit(R5_MadeGood, &dev->flags)) {
				rdev = conf->disks[i].rdev;
				rdev_clear_badblocks(rdev, sh->sector,
						     STRIPE_SECTORS);
				rdev_dec_pending(rdev, conf->mddev);
			}
3353 3354
		}

3355 3356 3357
	if (s.ops_request)
		raid_run_ops(sh, s.ops_request);

D
Dan Williams 已提交
3358
	ops_run_io(sh, &s);
3359

3360
	if (s.dec_preread_active) {
3361
		/* We delay this until after ops_run_io so that if make_request
T
Tejun Heo 已提交
3362
		 * is waiting on a flush, it won't continue until the writes
3363 3364 3365 3366 3367 3368 3369 3370
		 * have actually been submitted.
		 */
		atomic_dec(&conf->preread_active_stripes);
		if (atomic_read(&conf->preread_active_stripes) <
		    IO_THRESHOLD)
			md_wakeup_thread(conf->mddev->thread);
	}

3371
	return_io(s.return_bi);
3372

3373
	clear_bit(STRIPE_ACTIVE, &sh->state);
3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386
}

static void raid5_activate_delayed(raid5_conf_t *conf)
{
	if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD) {
		while (!list_empty(&conf->delayed_list)) {
			struct list_head *l = conf->delayed_list.next;
			struct stripe_head *sh;
			sh = list_entry(l, struct stripe_head, lru);
			list_del_init(l);
			clear_bit(STRIPE_DELAYED, &sh->state);
			if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
				atomic_inc(&conf->preread_active_stripes);
3387
			list_add_tail(&sh->lru, &conf->hold_list);
3388
		}
N
NeilBrown 已提交
3389
	}
3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405
}

static void activate_bit_delay(raid5_conf_t *conf)
{
	/* device_lock is held */
	struct list_head head;
	list_add(&head, &conf->bitmap_list);
	list_del_init(&conf->bitmap_list);
	while (!list_empty(&head)) {
		struct stripe_head *sh = list_entry(head.next, struct stripe_head, lru);
		list_del_init(&sh->lru);
		atomic_inc(&sh->count);
		__release_stripe(conf, sh);
	}
}

N
NeilBrown 已提交
3406
int md_raid5_congested(mddev_t *mddev, int bits)
3407
{
3408
	raid5_conf_t *conf = mddev->private;
3409 3410 3411 3412

	/* No difference between reads and writes.  Just check
	 * how busy the stripe_cache is
	 */
3413

3414 3415 3416 3417 3418 3419 3420 3421 3422
	if (conf->inactive_blocked)
		return 1;
	if (conf->quiesce)
		return 1;
	if (list_empty_careful(&conf->inactive_list))
		return 1;

	return 0;
}
N
NeilBrown 已提交
3423 3424 3425 3426 3427 3428 3429 3430 3431
EXPORT_SYMBOL_GPL(md_raid5_congested);

static int raid5_congested(void *data, int bits)
{
	mddev_t *mddev = data;

	return mddev_congested(mddev, bits) ||
		md_raid5_congested(mddev, bits);
}
3432

3433 3434 3435
/* We want read requests to align with chunks where possible,
 * but write requests don't need to.
 */
3436 3437 3438
static int raid5_mergeable_bvec(struct request_queue *q,
				struct bvec_merge_data *bvm,
				struct bio_vec *biovec)
3439 3440
{
	mddev_t *mddev = q->queuedata;
3441
	sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
3442
	int max;
3443
	unsigned int chunk_sectors = mddev->chunk_sectors;
3444
	unsigned int bio_sectors = bvm->bi_size >> 9;
3445

3446
	if ((bvm->bi_rw & 1) == WRITE)
3447 3448
		return biovec->bv_len; /* always allow writes to be mergeable */

3449 3450
	if (mddev->new_chunk_sectors < mddev->chunk_sectors)
		chunk_sectors = mddev->new_chunk_sectors;
3451 3452 3453 3454 3455 3456 3457 3458
	max =  (chunk_sectors - ((sector & (chunk_sectors - 1)) + bio_sectors)) << 9;
	if (max < 0) max = 0;
	if (max <= biovec->bv_len && bio_sectors == 0)
		return biovec->bv_len;
	else
		return max;
}

3459 3460 3461 3462

static int in_chunk_boundary(mddev_t *mddev, struct bio *bio)
{
	sector_t sector = bio->bi_sector + get_start_sect(bio->bi_bdev);
3463
	unsigned int chunk_sectors = mddev->chunk_sectors;
3464 3465
	unsigned int bio_sectors = bio->bi_size >> 9;

3466 3467
	if (mddev->new_chunk_sectors < mddev->chunk_sectors)
		chunk_sectors = mddev->new_chunk_sectors;
3468 3469 3470 3471
	return  chunk_sectors >=
		((sector & (chunk_sectors - 1)) + bio_sectors);
}

3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500
/*
 *  add bio to the retry LIFO  ( in O(1) ... we are in interrupt )
 *  later sampled by raid5d.
 */
static void add_bio_to_retry(struct bio *bi,raid5_conf_t *conf)
{
	unsigned long flags;

	spin_lock_irqsave(&conf->device_lock, flags);

	bi->bi_next = conf->retry_read_aligned_list;
	conf->retry_read_aligned_list = bi;

	spin_unlock_irqrestore(&conf->device_lock, flags);
	md_wakeup_thread(conf->mddev->thread);
}


static struct bio *remove_bio_from_retry(raid5_conf_t *conf)
{
	struct bio *bi;

	bi = conf->retry_read_aligned;
	if (bi) {
		conf->retry_read_aligned = NULL;
		return bi;
	}
	bi = conf->retry_read_aligned_list;
	if(bi) {
3501
		conf->retry_read_aligned_list = bi->bi_next;
3502
		bi->bi_next = NULL;
3503 3504 3505 3506
		/*
		 * this sets the active strip count to 1 and the processed
		 * strip count to zero (upper 8 bits)
		 */
3507 3508 3509 3510 3511 3512 3513
		bi->bi_phys_segments = 1; /* biased count of active stripes */
	}

	return bi;
}


3514 3515 3516 3517 3518 3519
/*
 *  The "raid5_align_endio" should check if the read succeeded and if it
 *  did, call bio_endio on the original bio (having bio_put the new bio
 *  first).
 *  If the read failed..
 */
3520
static void raid5_align_endio(struct bio *bi, int error)
3521 3522
{
	struct bio* raid_bi  = bi->bi_private;
3523 3524 3525 3526 3527
	mddev_t *mddev;
	raid5_conf_t *conf;
	int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags);
	mdk_rdev_t *rdev;

3528
	bio_put(bi);
3529 3530 3531

	rdev = (void*)raid_bi->bi_next;
	raid_bi->bi_next = NULL;
3532 3533
	mddev = rdev->mddev;
	conf = mddev->private;
3534 3535 3536 3537

	rdev_dec_pending(rdev, conf->mddev);

	if (!error && uptodate) {
3538
		bio_endio(raid_bi, 0);
3539 3540
		if (atomic_dec_and_test(&conf->active_aligned_reads))
			wake_up(&conf->wait_for_stripe);
3541
		return;
3542 3543 3544
	}


3545
	pr_debug("raid5_align_endio : io error...handing IO for a retry\n");
3546 3547

	add_bio_to_retry(raid_bi, conf);
3548 3549
}

3550 3551
static int bio_fits_rdev(struct bio *bi)
{
3552
	struct request_queue *q = bdev_get_queue(bi->bi_bdev);
3553

3554
	if ((bi->bi_size>>9) > queue_max_sectors(q))
3555 3556
		return 0;
	blk_recount_segments(q, bi);
3557
	if (bi->bi_phys_segments > queue_max_segments(q))
3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569
		return 0;

	if (q->merge_bvec_fn)
		/* it's too hard to apply the merge_bvec_fn at this stage,
		 * just just give up
		 */
		return 0;

	return 1;
}


3570
static int chunk_aligned_read(mddev_t *mddev, struct bio * raid_bio)
3571
{
3572
	raid5_conf_t *conf = mddev->private;
N
NeilBrown 已提交
3573
	int dd_idx;
3574 3575 3576 3577
	struct bio* align_bi;
	mdk_rdev_t *rdev;

	if (!in_chunk_boundary(mddev, raid_bio)) {
3578
		pr_debug("chunk_aligned_read : non aligned\n");
3579 3580 3581
		return 0;
	}
	/*
3582
	 * use bio_clone_mddev to make a copy of the bio
3583
	 */
3584
	align_bi = bio_clone_mddev(raid_bio, GFP_NOIO, mddev);
3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595
	if (!align_bi)
		return 0;
	/*
	 *   set bi_end_io to a new function, and set bi_private to the
	 *     original bio.
	 */
	align_bi->bi_end_io  = raid5_align_endio;
	align_bi->bi_private = raid_bio;
	/*
	 *	compute position
	 */
3596 3597
	align_bi->bi_sector =  raid5_compute_sector(conf, raid_bio->bi_sector,
						    0,
3598
						    &dd_idx, NULL);
3599 3600 3601 3602

	rcu_read_lock();
	rdev = rcu_dereference(conf->disks[dd_idx].rdev);
	if (rdev && test_bit(In_sync, &rdev->flags)) {
3603 3604 3605
		sector_t first_bad;
		int bad_sectors;

3606 3607
		atomic_inc(&rdev->nr_pending);
		rcu_read_unlock();
3608 3609 3610 3611 3612
		raid_bio->bi_next = (void*)rdev;
		align_bi->bi_bdev =  rdev->bdev;
		align_bi->bi_flags &= ~(1 << BIO_SEG_VALID);
		align_bi->bi_sector += rdev->data_offset;

3613 3614 3615 3616
		if (!bio_fits_rdev(align_bi) ||
		    is_badblock(rdev, align_bi->bi_sector, align_bi->bi_size>>9,
				&first_bad, &bad_sectors)) {
			/* too big in some way, or has a known bad block */
3617 3618 3619 3620 3621
			bio_put(align_bi);
			rdev_dec_pending(rdev, mddev);
			return 0;
		}

3622 3623 3624 3625 3626 3627 3628
		spin_lock_irq(&conf->device_lock);
		wait_event_lock_irq(conf->wait_for_stripe,
				    conf->quiesce == 0,
				    conf->device_lock, /* nothing */);
		atomic_inc(&conf->active_aligned_reads);
		spin_unlock_irq(&conf->device_lock);

3629 3630 3631 3632
		generic_make_request(align_bi);
		return 1;
	} else {
		rcu_read_unlock();
3633
		bio_put(align_bi);
3634 3635 3636 3637
		return 0;
	}
}

3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689
/* __get_priority_stripe - get the next stripe to process
 *
 * Full stripe writes are allowed to pass preread active stripes up until
 * the bypass_threshold is exceeded.  In general the bypass_count
 * increments when the handle_list is handled before the hold_list; however, it
 * will not be incremented when STRIPE_IO_STARTED is sampled set signifying a
 * stripe with in flight i/o.  The bypass_count will be reset when the
 * head of the hold_list has changed, i.e. the head was promoted to the
 * handle_list.
 */
static struct stripe_head *__get_priority_stripe(raid5_conf_t *conf)
{
	struct stripe_head *sh;

	pr_debug("%s: handle: %s hold: %s full_writes: %d bypass_count: %d\n",
		  __func__,
		  list_empty(&conf->handle_list) ? "empty" : "busy",
		  list_empty(&conf->hold_list) ? "empty" : "busy",
		  atomic_read(&conf->pending_full_writes), conf->bypass_count);

	if (!list_empty(&conf->handle_list)) {
		sh = list_entry(conf->handle_list.next, typeof(*sh), lru);

		if (list_empty(&conf->hold_list))
			conf->bypass_count = 0;
		else if (!test_bit(STRIPE_IO_STARTED, &sh->state)) {
			if (conf->hold_list.next == conf->last_hold)
				conf->bypass_count++;
			else {
				conf->last_hold = conf->hold_list.next;
				conf->bypass_count -= conf->bypass_threshold;
				if (conf->bypass_count < 0)
					conf->bypass_count = 0;
			}
		}
	} else if (!list_empty(&conf->hold_list) &&
		   ((conf->bypass_threshold &&
		     conf->bypass_count > conf->bypass_threshold) ||
		    atomic_read(&conf->pending_full_writes) == 0)) {
		sh = list_entry(conf->hold_list.next,
				typeof(*sh), lru);
		conf->bypass_count -= conf->bypass_threshold;
		if (conf->bypass_count < 0)
			conf->bypass_count = 0;
	} else
		return NULL;

	list_del_init(&sh->lru);
	atomic_inc(&sh->count);
	BUG_ON(atomic_read(&sh->count) != 1);
	return sh;
}
3690

3691
static int make_request(mddev_t *mddev, struct bio * bi)
L
Linus Torvalds 已提交
3692
{
3693
	raid5_conf_t *conf = mddev->private;
3694
	int dd_idx;
L
Linus Torvalds 已提交
3695 3696 3697
	sector_t new_sector;
	sector_t logical_sector, last_sector;
	struct stripe_head *sh;
3698
	const int rw = bio_data_dir(bi);
3699
	int remaining;
3700
	int plugged;
L
Linus Torvalds 已提交
3701

T
Tejun Heo 已提交
3702 3703
	if (unlikely(bi->bi_rw & REQ_FLUSH)) {
		md_flush_request(mddev, bi);
3704 3705 3706
		return 0;
	}

3707
	md_write_start(mddev, bi);
3708

3709
	if (rw == READ &&
3710
	     mddev->reshape_position == MaxSector &&
3711
	     chunk_aligned_read(mddev,bi))
3712
		return 0;
3713

L
Linus Torvalds 已提交
3714 3715 3716 3717
	logical_sector = bi->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
	last_sector = bi->bi_sector + (bi->bi_size>>9);
	bi->bi_next = NULL;
	bi->bi_phys_segments = 1;	/* over-loaded to count active stripes */
3718

3719
	plugged = mddev_check_plugged(mddev);
L
Linus Torvalds 已提交
3720 3721
	for (;logical_sector < last_sector; logical_sector += STRIPE_SECTORS) {
		DEFINE_WAIT(w);
3722
		int disks, data_disks;
3723
		int previous;
3724

3725
	retry:
3726
		previous = 0;
3727
		disks = conf->raid_disks;
3728
		prepare_to_wait(&conf->wait_for_overlap, &w, TASK_UNINTERRUPTIBLE);
3729
		if (unlikely(conf->reshape_progress != MaxSector)) {
3730
			/* spinlock is needed as reshape_progress may be
3731 3732
			 * 64bit on a 32bit platform, and so it might be
			 * possible to see a half-updated value
3733
			 * Of course reshape_progress could change after
3734 3735 3736 3737
			 * the lock is dropped, so once we get a reference
			 * to the stripe that we think it is, we will have
			 * to check again.
			 */
3738
			spin_lock_irq(&conf->device_lock);
3739 3740 3741
			if (mddev->delta_disks < 0
			    ? logical_sector < conf->reshape_progress
			    : logical_sector >= conf->reshape_progress) {
3742
				disks = conf->previous_raid_disks;
3743 3744
				previous = 1;
			} else {
3745 3746 3747
				if (mddev->delta_disks < 0
				    ? logical_sector < conf->reshape_safe
				    : logical_sector >= conf->reshape_safe) {
3748 3749 3750 3751 3752
					spin_unlock_irq(&conf->device_lock);
					schedule();
					goto retry;
				}
			}
3753 3754
			spin_unlock_irq(&conf->device_lock);
		}
3755 3756
		data_disks = disks - conf->max_degraded;

3757 3758
		new_sector = raid5_compute_sector(conf, logical_sector,
						  previous,
3759
						  &dd_idx, NULL);
3760
		pr_debug("raid456: make_request, sector %llu logical %llu\n",
L
Linus Torvalds 已提交
3761 3762 3763
			(unsigned long long)new_sector, 
			(unsigned long long)logical_sector);

3764
		sh = get_active_stripe(conf, new_sector, previous,
3765
				       (bi->bi_rw&RWA_MASK), 0);
L
Linus Torvalds 已提交
3766
		if (sh) {
3767
			if (unlikely(previous)) {
3768
				/* expansion might have moved on while waiting for a
3769 3770 3771 3772 3773 3774
				 * stripe, so we must do the range check again.
				 * Expansion could still move past after this
				 * test, but as we are holding a reference to
				 * 'sh', we know that if that happens,
				 *  STRIPE_EXPANDING will get set and the expansion
				 * won't proceed until we finish with the stripe.
3775 3776 3777
				 */
				int must_retry = 0;
				spin_lock_irq(&conf->device_lock);
3778 3779 3780
				if (mddev->delta_disks < 0
				    ? logical_sector >= conf->reshape_progress
				    : logical_sector < conf->reshape_progress)
3781 3782 3783 3784 3785
					/* mismatch, need to try again */
					must_retry = 1;
				spin_unlock_irq(&conf->device_lock);
				if (must_retry) {
					release_stripe(sh);
3786
					schedule();
3787 3788 3789
					goto retry;
				}
			}
3790

3791
			if (rw == WRITE &&
3792
			    logical_sector >= mddev->suspend_lo &&
3793 3794
			    logical_sector < mddev->suspend_hi) {
				release_stripe(sh);
3795 3796 3797 3798 3799 3800 3801 3802 3803 3804
				/* As the suspend_* range is controlled by
				 * userspace, we want an interruptible
				 * wait.
				 */
				flush_signals(current);
				prepare_to_wait(&conf->wait_for_overlap,
						&w, TASK_INTERRUPTIBLE);
				if (logical_sector >= mddev->suspend_lo &&
				    logical_sector < mddev->suspend_hi)
					schedule();
3805 3806
				goto retry;
			}
3807 3808

			if (test_bit(STRIPE_EXPANDING, &sh->state) ||
3809
			    !add_stripe_bio(sh, bi, dd_idx, rw)) {
3810 3811
				/* Stripe is busy expanding or
				 * add failed due to overlap.  Flush everything
L
Linus Torvalds 已提交
3812 3813
				 * and wait a while
				 */
N
NeilBrown 已提交
3814
				md_wakeup_thread(mddev->thread);
L
Linus Torvalds 已提交
3815 3816 3817 3818 3819
				release_stripe(sh);
				schedule();
				goto retry;
			}
			finish_wait(&conf->wait_for_overlap, &w);
3820 3821
			set_bit(STRIPE_HANDLE, &sh->state);
			clear_bit(STRIPE_DELAYED, &sh->state);
T
Tejun Heo 已提交
3822
			if ((bi->bi_rw & REQ_SYNC) &&
3823 3824
			    !test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
				atomic_inc(&conf->preread_active_stripes);
L
Linus Torvalds 已提交
3825 3826 3827 3828 3829 3830 3831 3832 3833
			release_stripe(sh);
		} else {
			/* cannot get stripe for read-ahead, just give-up */
			clear_bit(BIO_UPTODATE, &bi->bi_flags);
			finish_wait(&conf->wait_for_overlap, &w);
			break;
		}
			
	}
3834 3835 3836
	if (!plugged)
		md_wakeup_thread(mddev->thread);

L
Linus Torvalds 已提交
3837
	spin_lock_irq(&conf->device_lock);
3838
	remaining = raid5_dec_bi_phys_segments(bi);
3839 3840
	spin_unlock_irq(&conf->device_lock);
	if (remaining == 0) {
L
Linus Torvalds 已提交
3841

3842
		if ( rw == WRITE )
L
Linus Torvalds 已提交
3843
			md_write_end(mddev);
3844

3845
		bio_endio(bi, 0);
L
Linus Torvalds 已提交
3846
	}
3847

L
Linus Torvalds 已提交
3848 3849 3850
	return 0;
}

D
Dan Williams 已提交
3851 3852
static sector_t raid5_size(mddev_t *mddev, sector_t sectors, int raid_disks);

3853
static sector_t reshape_request(mddev_t *mddev, sector_t sector_nr, int *skipped)
L
Linus Torvalds 已提交
3854
{
3855 3856 3857 3858 3859 3860 3861 3862 3863
	/* reshaping is quite different to recovery/resync so it is
	 * handled quite separately ... here.
	 *
	 * On each call to sync_request, we gather one chunk worth of
	 * destination stripes and flag them as expanding.
	 * Then we find all the source stripes and request reads.
	 * As the reads complete, handle_stripe will copy the data
	 * into the destination stripe and release that stripe.
	 */
3864
	raid5_conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
3865
	struct stripe_head *sh;
3866
	sector_t first_sector, last_sector;
3867 3868 3869
	int raid_disks = conf->previous_raid_disks;
	int data_disks = raid_disks - conf->max_degraded;
	int new_data_disks = conf->raid_disks - conf->max_degraded;
3870 3871
	int i;
	int dd_idx;
3872
	sector_t writepos, readpos, safepos;
3873
	sector_t stripe_addr;
3874
	int reshape_sectors;
3875
	struct list_head stripes;
3876

3877 3878 3879 3880 3881 3882
	if (sector_nr == 0) {
		/* If restarting in the middle, skip the initial sectors */
		if (mddev->delta_disks < 0 &&
		    conf->reshape_progress < raid5_size(mddev, 0, 0)) {
			sector_nr = raid5_size(mddev, 0, 0)
				- conf->reshape_progress;
3883
		} else if (mddev->delta_disks >= 0 &&
3884 3885
			   conf->reshape_progress > 0)
			sector_nr = conf->reshape_progress;
3886
		sector_div(sector_nr, new_data_disks);
3887
		if (sector_nr) {
3888 3889
			mddev->curr_resync_completed = sector_nr;
			sysfs_notify(&mddev->kobj, NULL, "sync_completed");
3890 3891 3892
			*skipped = 1;
			return sector_nr;
		}
3893 3894
	}

3895 3896 3897 3898
	/* We need to process a full chunk at a time.
	 * If old and new chunk sizes differ, we need to process the
	 * largest of these
	 */
3899 3900
	if (mddev->new_chunk_sectors > mddev->chunk_sectors)
		reshape_sectors = mddev->new_chunk_sectors;
3901
	else
3902
		reshape_sectors = mddev->chunk_sectors;
3903

3904 3905 3906 3907 3908
	/* we update the metadata when there is more than 3Meg
	 * in the block range (that is rather arbitrary, should
	 * probably be time based) or when the data about to be
	 * copied would over-write the source of the data at
	 * the front of the range.
3909 3910
	 * i.e. one new_stripe along from reshape_progress new_maps
	 * to after where reshape_safe old_maps to
3911
	 */
3912
	writepos = conf->reshape_progress;
3913
	sector_div(writepos, new_data_disks);
3914 3915
	readpos = conf->reshape_progress;
	sector_div(readpos, data_disks);
3916
	safepos = conf->reshape_safe;
3917
	sector_div(safepos, data_disks);
3918
	if (mddev->delta_disks < 0) {
3919
		writepos -= min_t(sector_t, reshape_sectors, writepos);
3920
		readpos += reshape_sectors;
3921
		safepos += reshape_sectors;
3922
	} else {
3923
		writepos += reshape_sectors;
3924 3925
		readpos -= min_t(sector_t, reshape_sectors, readpos);
		safepos -= min_t(sector_t, reshape_sectors, safepos);
3926
	}
3927

3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944
	/* 'writepos' is the most advanced device address we might write.
	 * 'readpos' is the least advanced device address we might read.
	 * 'safepos' is the least address recorded in the metadata as having
	 *     been reshaped.
	 * If 'readpos' is behind 'writepos', then there is no way that we can
	 * ensure safety in the face of a crash - that must be done by userspace
	 * making a backup of the data.  So in that case there is no particular
	 * rush to update metadata.
	 * Otherwise if 'safepos' is behind 'writepos', then we really need to
	 * update the metadata to advance 'safepos' to match 'readpos' so that
	 * we can be safe in the event of a crash.
	 * So we insist on updating metadata if safepos is behind writepos and
	 * readpos is beyond writepos.
	 * In any case, update the metadata every 10 seconds.
	 * Maybe that number should be configurable, but I'm not sure it is
	 * worth it.... maybe it could be a multiple of safemode_delay???
	 */
3945
	if ((mddev->delta_disks < 0
3946 3947 3948
	     ? (safepos > writepos && readpos < writepos)
	     : (safepos < writepos && readpos > writepos)) ||
	    time_after(jiffies, conf->reshape_checkpoint + 10*HZ)) {
3949 3950 3951
		/* Cannot proceed until we've updated the superblock... */
		wait_event(conf->wait_for_overlap,
			   atomic_read(&conf->reshape_stripes)==0);
3952
		mddev->reshape_position = conf->reshape_progress;
3953
		mddev->curr_resync_completed = sector_nr;
3954
		conf->reshape_checkpoint = jiffies;
3955
		set_bit(MD_CHANGE_DEVS, &mddev->flags);
3956
		md_wakeup_thread(mddev->thread);
3957
		wait_event(mddev->sb_wait, mddev->flags == 0 ||
3958 3959
			   kthread_should_stop());
		spin_lock_irq(&conf->device_lock);
3960
		conf->reshape_safe = mddev->reshape_position;
3961 3962
		spin_unlock_irq(&conf->device_lock);
		wake_up(&conf->wait_for_overlap);
3963
		sysfs_notify(&mddev->kobj, NULL, "sync_completed");
3964 3965
	}

3966 3967 3968 3969
	if (mddev->delta_disks < 0) {
		BUG_ON(conf->reshape_progress == 0);
		stripe_addr = writepos;
		BUG_ON((mddev->dev_sectors &
3970 3971
			~((sector_t)reshape_sectors - 1))
		       - reshape_sectors - stripe_addr
3972 3973
		       != sector_nr);
	} else {
3974
		BUG_ON(writepos != sector_nr + reshape_sectors);
3975 3976
		stripe_addr = sector_nr;
	}
3977
	INIT_LIST_HEAD(&stripes);
3978
	for (i = 0; i < reshape_sectors; i += STRIPE_SECTORS) {
3979
		int j;
3980
		int skipped_disk = 0;
3981
		sh = get_active_stripe(conf, stripe_addr+i, 0, 0, 1);
3982 3983 3984 3985 3986 3987 3988 3989 3990
		set_bit(STRIPE_EXPANDING, &sh->state);
		atomic_inc(&conf->reshape_stripes);
		/* If any of this stripe is beyond the end of the old
		 * array, then we need to zero those blocks
		 */
		for (j=sh->disks; j--;) {
			sector_t s;
			if (j == sh->pd_idx)
				continue;
3991
			if (conf->level == 6 &&
3992
			    j == sh->qd_idx)
3993
				continue;
3994
			s = compute_blocknr(sh, j, 0);
D
Dan Williams 已提交
3995
			if (s < raid5_size(mddev, 0, 0)) {
3996
				skipped_disk = 1;
3997 3998 3999 4000 4001 4002
				continue;
			}
			memset(page_address(sh->dev[j].page), 0, STRIPE_SIZE);
			set_bit(R5_Expanded, &sh->dev[j].flags);
			set_bit(R5_UPTODATE, &sh->dev[j].flags);
		}
4003
		if (!skipped_disk) {
4004 4005 4006
			set_bit(STRIPE_EXPAND_READY, &sh->state);
			set_bit(STRIPE_HANDLE, &sh->state);
		}
4007
		list_add(&sh->lru, &stripes);
4008 4009
	}
	spin_lock_irq(&conf->device_lock);
4010
	if (mddev->delta_disks < 0)
4011
		conf->reshape_progress -= reshape_sectors * new_data_disks;
4012
	else
4013
		conf->reshape_progress += reshape_sectors * new_data_disks;
4014 4015 4016 4017 4018 4019 4020
	spin_unlock_irq(&conf->device_lock);
	/* Ok, those stripe are ready. We can start scheduling
	 * reads on the source stripes.
	 * The source stripes are determined by mapping the first and last
	 * block on the destination stripes.
	 */
	first_sector =
4021
		raid5_compute_sector(conf, stripe_addr*(new_data_disks),
4022
				     1, &dd_idx, NULL);
4023
	last_sector =
4024
		raid5_compute_sector(conf, ((stripe_addr+reshape_sectors)
4025
					    * new_data_disks - 1),
4026
				     1, &dd_idx, NULL);
A
Andre Noll 已提交
4027 4028
	if (last_sector >= mddev->dev_sectors)
		last_sector = mddev->dev_sectors - 1;
4029
	while (first_sector <= last_sector) {
4030
		sh = get_active_stripe(conf, first_sector, 1, 0, 1);
4031 4032 4033 4034 4035
		set_bit(STRIPE_EXPAND_SOURCE, &sh->state);
		set_bit(STRIPE_HANDLE, &sh->state);
		release_stripe(sh);
		first_sector += STRIPE_SECTORS;
	}
4036 4037 4038 4039 4040 4041 4042 4043
	/* Now that the sources are clearly marked, we can release
	 * the destination stripes
	 */
	while (!list_empty(&stripes)) {
		sh = list_entry(stripes.next, struct stripe_head, lru);
		list_del_init(&sh->lru);
		release_stripe(sh);
	}
4044 4045 4046
	/* If this takes us to the resync_max point where we have to pause,
	 * then we need to write out the superblock.
	 */
4047
	sector_nr += reshape_sectors;
4048 4049
	if ((sector_nr - mddev->curr_resync_completed) * 2
	    >= mddev->resync_max - mddev->curr_resync_completed) {
4050 4051 4052
		/* Cannot proceed until we've updated the superblock... */
		wait_event(conf->wait_for_overlap,
			   atomic_read(&conf->reshape_stripes) == 0);
4053
		mddev->reshape_position = conf->reshape_progress;
4054
		mddev->curr_resync_completed = sector_nr;
4055
		conf->reshape_checkpoint = jiffies;
4056 4057 4058 4059 4060 4061
		set_bit(MD_CHANGE_DEVS, &mddev->flags);
		md_wakeup_thread(mddev->thread);
		wait_event(mddev->sb_wait,
			   !test_bit(MD_CHANGE_DEVS, &mddev->flags)
			   || kthread_should_stop());
		spin_lock_irq(&conf->device_lock);
4062
		conf->reshape_safe = mddev->reshape_position;
4063 4064
		spin_unlock_irq(&conf->device_lock);
		wake_up(&conf->wait_for_overlap);
4065
		sysfs_notify(&mddev->kobj, NULL, "sync_completed");
4066
	}
4067
	return reshape_sectors;
4068 4069 4070 4071 4072
}

/* FIXME go_faster isn't used */
static inline sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster)
{
4073
	raid5_conf_t *conf = mddev->private;
4074
	struct stripe_head *sh;
A
Andre Noll 已提交
4075
	sector_t max_sector = mddev->dev_sectors;
N
NeilBrown 已提交
4076
	sector_t sync_blocks;
4077 4078
	int still_degraded = 0;
	int i;
L
Linus Torvalds 已提交
4079

4080
	if (sector_nr >= max_sector) {
L
Linus Torvalds 已提交
4081
		/* just being told to finish up .. nothing much to do */
4082

4083 4084 4085 4086
		if (test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery)) {
			end_reshape(conf);
			return 0;
		}
4087 4088 4089 4090

		if (mddev->curr_resync < max_sector) /* aborted */
			bitmap_end_sync(mddev->bitmap, mddev->curr_resync,
					&sync_blocks, 1);
4091
		else /* completed sync */
4092 4093 4094
			conf->fullsync = 0;
		bitmap_close_sync(mddev->bitmap);

L
Linus Torvalds 已提交
4095 4096
		return 0;
	}
4097

4098 4099 4100
	/* Allow raid5_quiesce to complete */
	wait_event(conf->wait_for_overlap, conf->quiesce != 2);

4101 4102
	if (test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery))
		return reshape_request(mddev, sector_nr, skipped);
4103

4104 4105 4106 4107 4108 4109
	/* No need to check resync_max as we never do more than one
	 * stripe, and as resync_max will always be on a chunk boundary,
	 * if the check in md_do_sync didn't fire, there is no chance
	 * of overstepping resync_max here
	 */

4110
	/* if there is too many failed drives and we are trying
L
Linus Torvalds 已提交
4111 4112 4113
	 * to resync, then assert that we are finished, because there is
	 * nothing we can do.
	 */
4114
	if (mddev->degraded >= conf->max_degraded &&
4115
	    test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) {
A
Andre Noll 已提交
4116
		sector_t rv = mddev->dev_sectors - sector_nr;
4117
		*skipped = 1;
L
Linus Torvalds 已提交
4118 4119
		return rv;
	}
4120
	if (!bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, 1) &&
4121
	    !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery) &&
4122 4123 4124 4125 4126 4127
	    !conf->fullsync && sync_blocks >= STRIPE_SECTORS) {
		/* we can skip this block, and probably more */
		sync_blocks /= STRIPE_SECTORS;
		*skipped = 1;
		return sync_blocks * STRIPE_SECTORS; /* keep things rounded to whole stripes */
	}
L
Linus Torvalds 已提交
4128

N
NeilBrown 已提交
4129 4130 4131

	bitmap_cond_end_sync(mddev->bitmap, sector_nr);

4132
	sh = get_active_stripe(conf, sector_nr, 0, 1, 0);
L
Linus Torvalds 已提交
4133
	if (sh == NULL) {
4134
		sh = get_active_stripe(conf, sector_nr, 0, 0, 0);
L
Linus Torvalds 已提交
4135
		/* make sure we don't swamp the stripe cache if someone else
4136
		 * is trying to get access
L
Linus Torvalds 已提交
4137
		 */
4138
		schedule_timeout_uninterruptible(1);
L
Linus Torvalds 已提交
4139
	}
4140 4141 4142 4143
	/* Need to check if array will still be degraded after recovery/resync
	 * We don't need to check the 'failed' flag as when that gets set,
	 * recovery aborts.
	 */
4144
	for (i = 0; i < conf->raid_disks; i++)
4145 4146 4147 4148 4149
		if (conf->disks[i].rdev == NULL)
			still_degraded = 1;

	bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, still_degraded);

4150
	set_bit(STRIPE_SYNC_REQUESTED, &sh->state);
L
Linus Torvalds 已提交
4151

4152
	handle_stripe(sh);
L
Linus Torvalds 已提交
4153 4154 4155 4156 4157
	release_stripe(sh);

	return STRIPE_SECTORS;
}

4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170
static int  retry_aligned_read(raid5_conf_t *conf, struct bio *raid_bio)
{
	/* We may not be able to submit a whole bio at once as there
	 * may not be enough stripe_heads available.
	 * We cannot pre-allocate enough stripe_heads as we may need
	 * more than exist in the cache (if we allow ever large chunks).
	 * So we do one stripe head at a time and record in
	 * ->bi_hw_segments how many have been done.
	 *
	 * We *know* that this entire raid_bio is in one chunk, so
	 * it will be only one 'dd_idx' and only need one call to raid5_compute_sector.
	 */
	struct stripe_head *sh;
4171
	int dd_idx;
4172 4173 4174 4175 4176 4177
	sector_t sector, logical_sector, last_sector;
	int scnt = 0;
	int remaining;
	int handled = 0;

	logical_sector = raid_bio->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
4178
	sector = raid5_compute_sector(conf, logical_sector,
4179
				      0, &dd_idx, NULL);
4180 4181 4182
	last_sector = raid_bio->bi_sector + (raid_bio->bi_size>>9);

	for (; logical_sector < last_sector;
4183 4184 4185
	     logical_sector += STRIPE_SECTORS,
		     sector += STRIPE_SECTORS,
		     scnt++) {
4186

4187
		if (scnt < raid5_bi_hw_segments(raid_bio))
4188 4189 4190
			/* already done this stripe */
			continue;

4191
		sh = get_active_stripe(conf, sector, 0, 1, 0);
4192 4193 4194

		if (!sh) {
			/* failed to get a stripe - must wait */
4195
			raid5_set_bi_hw_segments(raid_bio, scnt);
4196 4197 4198 4199 4200
			conf->retry_read_aligned = raid_bio;
			return handled;
		}

		set_bit(R5_ReadError, &sh->dev[dd_idx].flags);
4201 4202
		if (!add_stripe_bio(sh, raid_bio, dd_idx, 0)) {
			release_stripe(sh);
4203
			raid5_set_bi_hw_segments(raid_bio, scnt);
4204 4205 4206 4207
			conf->retry_read_aligned = raid_bio;
			return handled;
		}

4208
		handle_stripe(sh);
4209 4210 4211 4212
		release_stripe(sh);
		handled++;
	}
	spin_lock_irq(&conf->device_lock);
4213
	remaining = raid5_dec_bi_phys_segments(raid_bio);
4214
	spin_unlock_irq(&conf->device_lock);
4215 4216
	if (remaining == 0)
		bio_endio(raid_bio, 0);
4217 4218 4219 4220 4221 4222
	if (atomic_dec_and_test(&conf->active_aligned_reads))
		wake_up(&conf->wait_for_stripe);
	return handled;
}


L
Linus Torvalds 已提交
4223 4224 4225 4226 4227 4228 4229
/*
 * This is our raid5 kernel thread.
 *
 * We scan the hash table for stripes which can be handled now.
 * During the scan, completed stripes are saved for us by the interrupt
 * handler, so that they will not have to wait for our next wakeup.
 */
4230
static void raid5d(mddev_t *mddev)
L
Linus Torvalds 已提交
4231 4232
{
	struct stripe_head *sh;
4233
	raid5_conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
4234
	int handled;
4235
	struct blk_plug plug;
L
Linus Torvalds 已提交
4236

4237
	pr_debug("+++ raid5d active\n");
L
Linus Torvalds 已提交
4238 4239 4240

	md_check_recovery(mddev);

4241
	blk_start_plug(&plug);
L
Linus Torvalds 已提交
4242 4243 4244
	handled = 0;
	spin_lock_irq(&conf->device_lock);
	while (1) {
4245
		struct bio *bio;
L
Linus Torvalds 已提交
4246

4247 4248 4249 4250
		if (atomic_read(&mddev->plug_cnt) == 0 &&
		    !list_empty(&conf->bitmap_list)) {
			/* Now is a good time to flush some bitmap updates */
			conf->seq_flush++;
4251
			spin_unlock_irq(&conf->device_lock);
4252
			bitmap_unplug(mddev->bitmap);
4253
			spin_lock_irq(&conf->device_lock);
4254
			conf->seq_write = conf->seq_flush;
4255 4256
			activate_bit_delay(conf);
		}
4257 4258
		if (atomic_read(&mddev->plug_cnt) == 0)
			raid5_activate_delayed(conf);
4259

4260 4261 4262 4263 4264 4265 4266 4267 4268 4269
		while ((bio = remove_bio_from_retry(conf))) {
			int ok;
			spin_unlock_irq(&conf->device_lock);
			ok = retry_aligned_read(conf, bio);
			spin_lock_irq(&conf->device_lock);
			if (!ok)
				break;
			handled++;
		}

4270 4271
		sh = __get_priority_stripe(conf);

4272
		if (!sh)
L
Linus Torvalds 已提交
4273 4274 4275 4276
			break;
		spin_unlock_irq(&conf->device_lock);
		
		handled++;
4277 4278 4279
		handle_stripe(sh);
		release_stripe(sh);
		cond_resched();
L
Linus Torvalds 已提交
4280

4281 4282 4283
		if (mddev->flags & ~(1<<MD_CHANGE_PENDING))
			md_check_recovery(mddev);

L
Linus Torvalds 已提交
4284 4285
		spin_lock_irq(&conf->device_lock);
	}
4286
	pr_debug("%d stripes handled\n", handled);
L
Linus Torvalds 已提交
4287 4288 4289

	spin_unlock_irq(&conf->device_lock);

4290
	async_tx_issue_pending_all();
4291
	blk_finish_plug(&plug);
L
Linus Torvalds 已提交
4292

4293
	pr_debug("--- raid5d inactive\n");
L
Linus Torvalds 已提交
4294 4295
}

4296
static ssize_t
4297
raid5_show_stripe_cache_size(mddev_t *mddev, char *page)
4298
{
4299
	raid5_conf_t *conf = mddev->private;
4300 4301 4302 4303
	if (conf)
		return sprintf(page, "%d\n", conf->max_nr_stripes);
	else
		return 0;
4304 4305
}

4306 4307
int
raid5_set_cache_size(mddev_t *mddev, int size)
4308
{
4309
	raid5_conf_t *conf = mddev->private;
4310 4311
	int err;

4312
	if (size <= 16 || size > 32768)
4313
		return -EINVAL;
4314
	while (size < conf->max_nr_stripes) {
4315 4316 4317 4318 4319
		if (drop_one_stripe(conf))
			conf->max_nr_stripes--;
		else
			break;
	}
4320 4321 4322
	err = md_allow_write(mddev);
	if (err)
		return err;
4323
	while (size > conf->max_nr_stripes) {
4324 4325 4326 4327
		if (grow_one_stripe(conf))
			conf->max_nr_stripes++;
		else break;
	}
4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348
	return 0;
}
EXPORT_SYMBOL(raid5_set_cache_size);

static ssize_t
raid5_store_stripe_cache_size(mddev_t *mddev, const char *page, size_t len)
{
	raid5_conf_t *conf = mddev->private;
	unsigned long new;
	int err;

	if (len >= PAGE_SIZE)
		return -EINVAL;
	if (!conf)
		return -ENODEV;

	if (strict_strtoul(page, 10, &new))
		return -EINVAL;
	err = raid5_set_cache_size(mddev, new);
	if (err)
		return err;
4349 4350
	return len;
}
4351

4352 4353 4354 4355
static struct md_sysfs_entry
raid5_stripecache_size = __ATTR(stripe_cache_size, S_IRUGO | S_IWUSR,
				raid5_show_stripe_cache_size,
				raid5_store_stripe_cache_size);
4356

4357 4358 4359
static ssize_t
raid5_show_preread_threshold(mddev_t *mddev, char *page)
{
4360
	raid5_conf_t *conf = mddev->private;
4361 4362 4363 4364 4365 4366 4367 4368 4369
	if (conf)
		return sprintf(page, "%d\n", conf->bypass_threshold);
	else
		return 0;
}

static ssize_t
raid5_store_preread_threshold(mddev_t *mddev, const char *page, size_t len)
{
4370
	raid5_conf_t *conf = mddev->private;
4371
	unsigned long new;
4372 4373 4374 4375 4376
	if (len >= PAGE_SIZE)
		return -EINVAL;
	if (!conf)
		return -ENODEV;

4377
	if (strict_strtoul(page, 10, &new))
4378
		return -EINVAL;
4379
	if (new > conf->max_nr_stripes)
4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390
		return -EINVAL;
	conf->bypass_threshold = new;
	return len;
}

static struct md_sysfs_entry
raid5_preread_bypass_threshold = __ATTR(preread_bypass_threshold,
					S_IRUGO | S_IWUSR,
					raid5_show_preread_threshold,
					raid5_store_preread_threshold);

4391
static ssize_t
4392
stripe_cache_active_show(mddev_t *mddev, char *page)
4393
{
4394
	raid5_conf_t *conf = mddev->private;
4395 4396 4397 4398
	if (conf)
		return sprintf(page, "%d\n", atomic_read(&conf->active_stripes));
	else
		return 0;
4399 4400
}

4401 4402
static struct md_sysfs_entry
raid5_stripecache_active = __ATTR_RO(stripe_cache_active);
4403

4404
static struct attribute *raid5_attrs[] =  {
4405 4406
	&raid5_stripecache_size.attr,
	&raid5_stripecache_active.attr,
4407
	&raid5_preread_bypass_threshold.attr,
4408 4409
	NULL,
};
4410 4411 4412
static struct attribute_group raid5_attrs_group = {
	.name = NULL,
	.attrs = raid5_attrs,
4413 4414
};

4415 4416 4417
static sector_t
raid5_size(mddev_t *mddev, sector_t sectors, int raid_disks)
{
4418
	raid5_conf_t *conf = mddev->private;
4419 4420 4421

	if (!sectors)
		sectors = mddev->dev_sectors;
4422
	if (!raid_disks)
4423
		/* size is defined by the smallest of previous and new size */
4424
		raid_disks = min(conf->raid_disks, conf->previous_raid_disks);
4425

4426
	sectors &= ~((sector_t)mddev->chunk_sectors - 1);
4427
	sectors &= ~((sector_t)mddev->new_chunk_sectors - 1);
4428 4429 4430
	return sectors * (raid_disks - conf->max_degraded);
}

4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442
static void raid5_free_percpu(raid5_conf_t *conf)
{
	struct raid5_percpu *percpu;
	unsigned long cpu;

	if (!conf->percpu)
		return;

	get_online_cpus();
	for_each_possible_cpu(cpu) {
		percpu = per_cpu_ptr(conf->percpu, cpu);
		safe_put_page(percpu->spare_page);
4443
		kfree(percpu->scribble);
4444 4445 4446 4447 4448 4449 4450 4451 4452
	}
#ifdef CONFIG_HOTPLUG_CPU
	unregister_cpu_notifier(&conf->cpu_notify);
#endif
	put_online_cpus();

	free_percpu(conf->percpu);
}

4453 4454 4455
static void free_conf(raid5_conf_t *conf)
{
	shrink_stripes(conf);
4456
	raid5_free_percpu(conf);
4457 4458 4459 4460 4461
	kfree(conf->disks);
	kfree(conf->stripe_hashtbl);
	kfree(conf);
}

4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472
#ifdef CONFIG_HOTPLUG_CPU
static int raid456_cpu_notify(struct notifier_block *nfb, unsigned long action,
			      void *hcpu)
{
	raid5_conf_t *conf = container_of(nfb, raid5_conf_t, cpu_notify);
	long cpu = (long)hcpu;
	struct raid5_percpu *percpu = per_cpu_ptr(conf->percpu, cpu);

	switch (action) {
	case CPU_UP_PREPARE:
	case CPU_UP_PREPARE_FROZEN:
4473
		if (conf->level == 6 && !percpu->spare_page)
4474
			percpu->spare_page = alloc_page(GFP_KERNEL);
4475 4476 4477 4478 4479 4480 4481
		if (!percpu->scribble)
			percpu->scribble = kmalloc(conf->scribble_len, GFP_KERNEL);

		if (!percpu->scribble ||
		    (conf->level == 6 && !percpu->spare_page)) {
			safe_put_page(percpu->spare_page);
			kfree(percpu->scribble);
4482 4483
			pr_err("%s: failed memory allocation for cpu%ld\n",
			       __func__, cpu);
4484
			return notifier_from_errno(-ENOMEM);
4485 4486 4487 4488 4489
		}
		break;
	case CPU_DEAD:
	case CPU_DEAD_FROZEN:
		safe_put_page(percpu->spare_page);
4490
		kfree(percpu->scribble);
4491
		percpu->spare_page = NULL;
4492
		percpu->scribble = NULL;
4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504
		break;
	default:
		break;
	}
	return NOTIFY_OK;
}
#endif

static int raid5_alloc_percpu(raid5_conf_t *conf)
{
	unsigned long cpu;
	struct page *spare_page;
4505
	struct raid5_percpu __percpu *allcpus;
4506
	void *scribble;
4507 4508 4509 4510 4511 4512 4513 4514 4515 4516
	int err;

	allcpus = alloc_percpu(struct raid5_percpu);
	if (!allcpus)
		return -ENOMEM;
	conf->percpu = allcpus;

	get_online_cpus();
	err = 0;
	for_each_present_cpu(cpu) {
4517 4518 4519 4520 4521 4522 4523 4524
		if (conf->level == 6) {
			spare_page = alloc_page(GFP_KERNEL);
			if (!spare_page) {
				err = -ENOMEM;
				break;
			}
			per_cpu_ptr(conf->percpu, cpu)->spare_page = spare_page;
		}
4525
		scribble = kmalloc(conf->scribble_len, GFP_KERNEL);
4526
		if (!scribble) {
4527 4528 4529
			err = -ENOMEM;
			break;
		}
4530
		per_cpu_ptr(conf->percpu, cpu)->scribble = scribble;
4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542
	}
#ifdef CONFIG_HOTPLUG_CPU
	conf->cpu_notify.notifier_call = raid456_cpu_notify;
	conf->cpu_notify.priority = 0;
	if (err == 0)
		err = register_cpu_notifier(&conf->cpu_notify);
#endif
	put_online_cpus();

	return err;
}

N
NeilBrown 已提交
4543
static raid5_conf_t *setup_conf(mddev_t *mddev)
L
Linus Torvalds 已提交
4544 4545
{
	raid5_conf_t *conf;
4546
	int raid_disk, memory, max_disks;
L
Linus Torvalds 已提交
4547 4548 4549
	mdk_rdev_t *rdev;
	struct disk_info *disk;

N
NeilBrown 已提交
4550 4551 4552
	if (mddev->new_level != 5
	    && mddev->new_level != 4
	    && mddev->new_level != 6) {
4553
		printk(KERN_ERR "md/raid:%s: raid level not set to 4/5/6 (%d)\n",
N
NeilBrown 已提交
4554 4555
		       mdname(mddev), mddev->new_level);
		return ERR_PTR(-EIO);
L
Linus Torvalds 已提交
4556
	}
N
NeilBrown 已提交
4557 4558 4559 4560
	if ((mddev->new_level == 5
	     && !algorithm_valid_raid5(mddev->new_layout)) ||
	    (mddev->new_level == 6
	     && !algorithm_valid_raid6(mddev->new_layout))) {
4561
		printk(KERN_ERR "md/raid:%s: layout %d not supported\n",
N
NeilBrown 已提交
4562 4563
		       mdname(mddev), mddev->new_layout);
		return ERR_PTR(-EIO);
4564
	}
N
NeilBrown 已提交
4565
	if (mddev->new_level == 6 && mddev->raid_disks < 4) {
4566
		printk(KERN_ERR "md/raid:%s: not enough configured devices (%d, minimum 4)\n",
N
NeilBrown 已提交
4567 4568
		       mdname(mddev), mddev->raid_disks);
		return ERR_PTR(-EINVAL);
4569 4570
	}

4571 4572 4573
	if (!mddev->new_chunk_sectors ||
	    (mddev->new_chunk_sectors << 9) % PAGE_SIZE ||
	    !is_power_of_2(mddev->new_chunk_sectors)) {
4574 4575
		printk(KERN_ERR "md/raid:%s: invalid chunk size %d\n",
		       mdname(mddev), mddev->new_chunk_sectors << 9);
N
NeilBrown 已提交
4576
		return ERR_PTR(-EINVAL);
4577 4578
	}

N
NeilBrown 已提交
4579 4580
	conf = kzalloc(sizeof(raid5_conf_t), GFP_KERNEL);
	if (conf == NULL)
L
Linus Torvalds 已提交
4581
		goto abort;
4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593
	spin_lock_init(&conf->device_lock);
	init_waitqueue_head(&conf->wait_for_stripe);
	init_waitqueue_head(&conf->wait_for_overlap);
	INIT_LIST_HEAD(&conf->handle_list);
	INIT_LIST_HEAD(&conf->hold_list);
	INIT_LIST_HEAD(&conf->delayed_list);
	INIT_LIST_HEAD(&conf->bitmap_list);
	INIT_LIST_HEAD(&conf->inactive_list);
	atomic_set(&conf->active_stripes, 0);
	atomic_set(&conf->preread_active_stripes, 0);
	atomic_set(&conf->active_aligned_reads, 0);
	conf->bypass_threshold = BYPASS_THRESHOLD;
N
NeilBrown 已提交
4594 4595 4596 4597 4598

	conf->raid_disks = mddev->raid_disks;
	if (mddev->reshape_position == MaxSector)
		conf->previous_raid_disks = mddev->raid_disks;
	else
4599
		conf->previous_raid_disks = mddev->raid_disks - mddev->delta_disks;
4600 4601
	max_disks = max(conf->raid_disks, conf->previous_raid_disks);
	conf->scribble_len = scribble_len(max_disks);
4602

4603
	conf->disks = kzalloc(max_disks * sizeof(struct disk_info),
4604 4605 4606
			      GFP_KERNEL);
	if (!conf->disks)
		goto abort;
4607

L
Linus Torvalds 已提交
4608 4609
	conf->mddev = mddev;

4610
	if ((conf->stripe_hashtbl = kzalloc(PAGE_SIZE, GFP_KERNEL)) == NULL)
L
Linus Torvalds 已提交
4611 4612
		goto abort;

4613 4614 4615 4616
	conf->level = mddev->new_level;
	if (raid5_alloc_percpu(conf) != 0)
		goto abort;

4617
	pr_debug("raid456: run(%s) called.\n", mdname(mddev));
L
Linus Torvalds 已提交
4618

4619
	list_for_each_entry(rdev, &mddev->disks, same_set) {
L
Linus Torvalds 已提交
4620
		raid_disk = rdev->raid_disk;
4621
		if (raid_disk >= max_disks
L
Linus Torvalds 已提交
4622 4623 4624 4625 4626 4627
		    || raid_disk < 0)
			continue;
		disk = conf->disks + raid_disk;

		disk->rdev = rdev;

4628
		if (test_bit(In_sync, &rdev->flags)) {
L
Linus Torvalds 已提交
4629
			char b[BDEVNAME_SIZE];
4630 4631 4632
			printk(KERN_INFO "md/raid:%s: device %s operational as raid"
			       " disk %d\n",
			       mdname(mddev), bdevname(rdev->bdev, b), raid_disk);
J
Jonathan Brassow 已提交
4633
		} else if (rdev->saved_raid_disk != raid_disk)
4634 4635
			/* Cannot rely on bitmap to complete recovery */
			conf->fullsync = 1;
L
Linus Torvalds 已提交
4636 4637
	}

4638
	conf->chunk_sectors = mddev->new_chunk_sectors;
N
NeilBrown 已提交
4639
	conf->level = mddev->new_level;
4640 4641 4642 4643
	if (conf->level == 6)
		conf->max_degraded = 2;
	else
		conf->max_degraded = 1;
N
NeilBrown 已提交
4644
	conf->algorithm = mddev->new_layout;
L
Linus Torvalds 已提交
4645
	conf->max_nr_stripes = NR_STRIPES;
4646
	conf->reshape_progress = mddev->reshape_position;
4647
	if (conf->reshape_progress != MaxSector) {
4648
		conf->prev_chunk_sectors = mddev->chunk_sectors;
4649 4650
		conf->prev_algo = mddev->layout;
	}
L
Linus Torvalds 已提交
4651

N
NeilBrown 已提交
4652
	memory = conf->max_nr_stripes * (sizeof(struct stripe_head) +
4653
		 max_disks * ((sizeof(struct bio) + PAGE_SIZE))) / 1024;
N
NeilBrown 已提交
4654 4655
	if (grow_stripes(conf, conf->max_nr_stripes)) {
		printk(KERN_ERR
4656 4657
		       "md/raid:%s: couldn't allocate %dkB for buffers\n",
		       mdname(mddev), memory);
N
NeilBrown 已提交
4658 4659
		goto abort;
	} else
4660 4661
		printk(KERN_INFO "md/raid:%s: allocated %dkB\n",
		       mdname(mddev), memory);
L
Linus Torvalds 已提交
4662

4663
	conf->thread = md_register_thread(raid5d, mddev, NULL);
N
NeilBrown 已提交
4664 4665
	if (!conf->thread) {
		printk(KERN_ERR
4666
		       "md/raid:%s: couldn't allocate thread.\n",
N
NeilBrown 已提交
4667
		       mdname(mddev));
4668 4669
		goto abort;
	}
N
NeilBrown 已提交
4670 4671 4672 4673 4674

	return conf;

 abort:
	if (conf) {
4675
		free_conf(conf);
N
NeilBrown 已提交
4676 4677 4678 4679 4680
		return ERR_PTR(-EIO);
	} else
		return ERR_PTR(-ENOMEM);
}

4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707

static int only_parity(int raid_disk, int algo, int raid_disks, int max_degraded)
{
	switch (algo) {
	case ALGORITHM_PARITY_0:
		if (raid_disk < max_degraded)
			return 1;
		break;
	case ALGORITHM_PARITY_N:
		if (raid_disk >= raid_disks - max_degraded)
			return 1;
		break;
	case ALGORITHM_PARITY_0_6:
		if (raid_disk == 0 || 
		    raid_disk == raid_disks - 1)
			return 1;
		break;
	case ALGORITHM_LEFT_ASYMMETRIC_6:
	case ALGORITHM_RIGHT_ASYMMETRIC_6:
	case ALGORITHM_LEFT_SYMMETRIC_6:
	case ALGORITHM_RIGHT_SYMMETRIC_6:
		if (raid_disk == raid_disks - 1)
			return 1;
	}
	return 0;
}

N
NeilBrown 已提交
4708 4709 4710
static int run(mddev_t *mddev)
{
	raid5_conf_t *conf;
4711
	int working_disks = 0;
4712
	int dirty_parity_disks = 0;
N
NeilBrown 已提交
4713
	mdk_rdev_t *rdev;
4714
	sector_t reshape_offset = 0;
N
NeilBrown 已提交
4715

4716
	if (mddev->recovery_cp != MaxSector)
4717
		printk(KERN_NOTICE "md/raid:%s: not clean"
4718 4719
		       " -- starting background reconstruction\n",
		       mdname(mddev));
N
NeilBrown 已提交
4720 4721 4722 4723 4724 4725 4726 4727
	if (mddev->reshape_position != MaxSector) {
		/* Check that we can continue the reshape.
		 * Currently only disks can change, it must
		 * increase, and we must be past the point where
		 * a stripe over-writes itself
		 */
		sector_t here_new, here_old;
		int old_disks;
4728
		int max_degraded = (mddev->level == 6 ? 2 : 1);
N
NeilBrown 已提交
4729

4730
		if (mddev->new_level != mddev->level) {
4731
			printk(KERN_ERR "md/raid:%s: unsupported reshape "
N
NeilBrown 已提交
4732 4733 4734 4735 4736 4737 4738 4739 4740 4741
			       "required - aborting.\n",
			       mdname(mddev));
			return -EINVAL;
		}
		old_disks = mddev->raid_disks - mddev->delta_disks;
		/* reshape_position must be on a new-stripe boundary, and one
		 * further up in new geometry must map after here in old
		 * geometry.
		 */
		here_new = mddev->reshape_position;
4742
		if (sector_div(here_new, mddev->new_chunk_sectors *
N
NeilBrown 已提交
4743
			       (mddev->raid_disks - max_degraded))) {
4744 4745
			printk(KERN_ERR "md/raid:%s: reshape_position not "
			       "on a stripe boundary\n", mdname(mddev));
N
NeilBrown 已提交
4746 4747
			return -EINVAL;
		}
4748
		reshape_offset = here_new * mddev->new_chunk_sectors;
N
NeilBrown 已提交
4749 4750
		/* here_new is the stripe we will write to */
		here_old = mddev->reshape_position;
4751
		sector_div(here_old, mddev->chunk_sectors *
N
NeilBrown 已提交
4752 4753 4754
			   (old_disks-max_degraded));
		/* here_old is the first stripe that we might need to read
		 * from */
4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765
		if (mddev->delta_disks == 0) {
			/* We cannot be sure it is safe to start an in-place
			 * reshape.  It is only safe if user-space if monitoring
			 * and taking constant backups.
			 * mdadm always starts a situation like this in
			 * readonly mode so it can take control before
			 * allowing any writes.  So just check for that.
			 */
			if ((here_new * mddev->new_chunk_sectors != 
			     here_old * mddev->chunk_sectors) ||
			    mddev->ro == 0) {
4766 4767 4768
				printk(KERN_ERR "md/raid:%s: in-place reshape must be started"
				       " in read-only mode - aborting\n",
				       mdname(mddev));
4769 4770 4771 4772 4773 4774 4775
				return -EINVAL;
			}
		} else if (mddev->delta_disks < 0
		    ? (here_new * mddev->new_chunk_sectors <=
		       here_old * mddev->chunk_sectors)
		    : (here_new * mddev->new_chunk_sectors >=
		       here_old * mddev->chunk_sectors)) {
N
NeilBrown 已提交
4776
			/* Reading from the same stripe as writing to - bad */
4777 4778 4779
			printk(KERN_ERR "md/raid:%s: reshape_position too early for "
			       "auto-recovery - aborting.\n",
			       mdname(mddev));
N
NeilBrown 已提交
4780 4781
			return -EINVAL;
		}
4782 4783
		printk(KERN_INFO "md/raid:%s: reshape will continue\n",
		       mdname(mddev));
N
NeilBrown 已提交
4784 4785 4786 4787
		/* OK, we should be able to continue; */
	} else {
		BUG_ON(mddev->level != mddev->new_level);
		BUG_ON(mddev->layout != mddev->new_layout);
4788
		BUG_ON(mddev->chunk_sectors != mddev->new_chunk_sectors);
N
NeilBrown 已提交
4789
		BUG_ON(mddev->delta_disks != 0);
L
Linus Torvalds 已提交
4790
	}
N
NeilBrown 已提交
4791

4792 4793 4794 4795 4796
	if (mddev->private == NULL)
		conf = setup_conf(mddev);
	else
		conf = mddev->private;

N
NeilBrown 已提交
4797 4798 4799 4800 4801 4802 4803 4804 4805 4806
	if (IS_ERR(conf))
		return PTR_ERR(conf);

	mddev->thread = conf->thread;
	conf->thread = NULL;
	mddev->private = conf;

	/*
	 * 0 for a fully functional array, 1 or 2 for a degraded array.
	 */
4807 4808 4809
	list_for_each_entry(rdev, &mddev->disks, same_set) {
		if (rdev->raid_disk < 0)
			continue;
4810
		if (test_bit(In_sync, &rdev->flags)) {
N
NeilBrown 已提交
4811
			working_disks++;
4812 4813
			continue;
		}
4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841
		/* This disc is not fully in-sync.  However if it
		 * just stored parity (beyond the recovery_offset),
		 * when we don't need to be concerned about the
		 * array being dirty.
		 * When reshape goes 'backwards', we never have
		 * partially completed devices, so we only need
		 * to worry about reshape going forwards.
		 */
		/* Hack because v0.91 doesn't store recovery_offset properly. */
		if (mddev->major_version == 0 &&
		    mddev->minor_version > 90)
			rdev->recovery_offset = reshape_offset;
			
		if (rdev->recovery_offset < reshape_offset) {
			/* We need to check old and new layout */
			if (!only_parity(rdev->raid_disk,
					 conf->algorithm,
					 conf->raid_disks,
					 conf->max_degraded))
				continue;
		}
		if (!only_parity(rdev->raid_disk,
				 conf->prev_algo,
				 conf->previous_raid_disks,
				 conf->max_degraded))
			continue;
		dirty_parity_disks++;
	}
N
NeilBrown 已提交
4842

4843 4844
	mddev->degraded = (max(conf->raid_disks, conf->previous_raid_disks)
			   - working_disks);
N
NeilBrown 已提交
4845

4846
	if (has_failed(conf)) {
4847
		printk(KERN_ERR "md/raid:%s: not enough operational devices"
L
Linus Torvalds 已提交
4848
			" (%d/%d failed)\n",
4849
			mdname(mddev), mddev->degraded, conf->raid_disks);
L
Linus Torvalds 已提交
4850 4851 4852
		goto abort;
	}

N
NeilBrown 已提交
4853
	/* device size must be a multiple of chunk size */
4854
	mddev->dev_sectors &= ~(mddev->chunk_sectors - 1);
N
NeilBrown 已提交
4855 4856
	mddev->resync_max_sectors = mddev->dev_sectors;

4857
	if (mddev->degraded > dirty_parity_disks &&
L
Linus Torvalds 已提交
4858
	    mddev->recovery_cp != MaxSector) {
4859 4860
		if (mddev->ok_start_degraded)
			printk(KERN_WARNING
4861 4862
			       "md/raid:%s: starting dirty degraded array"
			       " - data corruption possible.\n",
4863 4864 4865
			       mdname(mddev));
		else {
			printk(KERN_ERR
4866
			       "md/raid:%s: cannot start dirty degraded array.\n",
4867 4868 4869
			       mdname(mddev));
			goto abort;
		}
L
Linus Torvalds 已提交
4870 4871 4872
	}

	if (mddev->degraded == 0)
4873 4874
		printk(KERN_INFO "md/raid:%s: raid level %d active with %d out of %d"
		       " devices, algorithm %d\n", mdname(mddev), conf->level,
4875 4876
		       mddev->raid_disks-mddev->degraded, mddev->raid_disks,
		       mddev->new_layout);
L
Linus Torvalds 已提交
4877
	else
4878 4879 4880 4881 4882
		printk(KERN_ALERT "md/raid:%s: raid level %d active with %d"
		       " out of %d devices, algorithm %d\n",
		       mdname(mddev), conf->level,
		       mddev->raid_disks - mddev->degraded,
		       mddev->raid_disks, mddev->new_layout);
L
Linus Torvalds 已提交
4883 4884 4885

	print_raid5_conf(conf);

4886 4887
	if (conf->reshape_progress != MaxSector) {
		conf->reshape_safe = conf->reshape_progress;
4888 4889 4890 4891 4892 4893
		atomic_set(&conf->reshape_stripes, 0);
		clear_bit(MD_RECOVERY_SYNC, &mddev->recovery);
		clear_bit(MD_RECOVERY_CHECK, &mddev->recovery);
		set_bit(MD_RECOVERY_RESHAPE, &mddev->recovery);
		set_bit(MD_RECOVERY_RUNNING, &mddev->recovery);
		mddev->sync_thread = md_register_thread(md_do_sync, mddev,
4894
							"reshape");
4895 4896
	}

L
Linus Torvalds 已提交
4897 4898

	/* Ok, everything is just fine now */
4899 4900
	if (mddev->to_remove == &raid5_attrs_group)
		mddev->to_remove = NULL;
N
NeilBrown 已提交
4901 4902
	else if (mddev->kobj.sd &&
	    sysfs_create_group(&mddev->kobj, &raid5_attrs_group))
4903
		printk(KERN_WARNING
4904
		       "raid5: failed to create sysfs attributes for %s\n",
4905
		       mdname(mddev));
4906
	md_set_array_sectors(mddev, raid5_size(mddev, 0, 0));
4907

4908
	if (mddev->queue) {
4909
		int chunk_size;
4910 4911 4912 4913 4914 4915 4916 4917 4918
		/* read-ahead size must cover two whole stripes, which
		 * is 2 * (datadisks) * chunksize where 'n' is the
		 * number of raid devices
		 */
		int data_disks = conf->previous_raid_disks - conf->max_degraded;
		int stripe = data_disks *
			((mddev->chunk_sectors << 9) / PAGE_SIZE);
		if (mddev->queue->backing_dev_info.ra_pages < 2 * stripe)
			mddev->queue->backing_dev_info.ra_pages = 2 * stripe;
N
NeilBrown 已提交
4919

4920
		blk_queue_merge_bvec(mddev->queue, raid5_mergeable_bvec);
4921

N
NeilBrown 已提交
4922 4923
		mddev->queue->backing_dev_info.congested_data = mddev;
		mddev->queue->backing_dev_info.congested_fn = raid5_congested;
4924

4925 4926 4927 4928
		chunk_size = mddev->chunk_sectors << 9;
		blk_queue_io_min(mddev->queue, chunk_size);
		blk_queue_io_opt(mddev->queue, chunk_size *
				 (conf->raid_disks - conf->max_degraded));
4929

4930 4931 4932 4933
		list_for_each_entry(rdev, &mddev->disks, same_set)
			disk_stack_limits(mddev->gendisk, rdev->bdev,
					  rdev->data_offset << 9);
	}
4934

L
Linus Torvalds 已提交
4935 4936
	return 0;
abort:
4937
	md_unregister_thread(&mddev->thread);
N
NeilBrown 已提交
4938 4939
	print_raid5_conf(conf);
	free_conf(conf);
L
Linus Torvalds 已提交
4940
	mddev->private = NULL;
4941
	printk(KERN_ALERT "md/raid:%s: failed to run raid set.\n", mdname(mddev));
L
Linus Torvalds 已提交
4942 4943 4944
	return -EIO;
}

4945
static int stop(mddev_t *mddev)
L
Linus Torvalds 已提交
4946
{
4947
	raid5_conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
4948

4949
	md_unregister_thread(&mddev->thread);
N
NeilBrown 已提交
4950 4951
	if (mddev->queue)
		mddev->queue->backing_dev_info.congested_fn = NULL;
4952
	free_conf(conf);
4953 4954
	mddev->private = NULL;
	mddev->to_remove = &raid5_attrs_group;
L
Linus Torvalds 已提交
4955 4956 4957
	return 0;
}

4958
static void status(struct seq_file *seq, mddev_t *mddev)
L
Linus Torvalds 已提交
4959
{
4960
	raid5_conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
4961 4962
	int i;

4963 4964
	seq_printf(seq, " level %d, %dk chunk, algorithm %d", mddev->level,
		mddev->chunk_sectors / 2, mddev->layout);
4965
	seq_printf (seq, " [%d/%d] [", conf->raid_disks, conf->raid_disks - mddev->degraded);
L
Linus Torvalds 已提交
4966 4967 4968
	for (i = 0; i < conf->raid_disks; i++)
		seq_printf (seq, "%s",
			       conf->disks[i].rdev &&
4969
			       test_bit(In_sync, &conf->disks[i].rdev->flags) ? "U" : "_");
L
Linus Torvalds 已提交
4970 4971 4972 4973 4974 4975 4976 4977
	seq_printf (seq, "]");
}

static void print_raid5_conf (raid5_conf_t *conf)
{
	int i;
	struct disk_info *tmp;

4978
	printk(KERN_DEBUG "RAID conf printout:\n");
L
Linus Torvalds 已提交
4979 4980 4981 4982
	if (!conf) {
		printk("(conf==NULL)\n");
		return;
	}
4983 4984 4985
	printk(KERN_DEBUG " --- level:%d rd:%d wd:%d\n", conf->level,
	       conf->raid_disks,
	       conf->raid_disks - conf->mddev->degraded);
L
Linus Torvalds 已提交
4986 4987 4988 4989 4990

	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
		tmp = conf->disks + i;
		if (tmp->rdev)
4991 4992 4993
			printk(KERN_DEBUG " disk %d, o:%d, dev:%s\n",
			       i, !test_bit(Faulty, &tmp->rdev->flags),
			       bdevname(tmp->rdev->bdev, b));
L
Linus Torvalds 已提交
4994 4995 4996 4997 4998 4999 5000 5001
	}
}

static int raid5_spare_active(mddev_t *mddev)
{
	int i;
	raid5_conf_t *conf = mddev->private;
	struct disk_info *tmp;
5002 5003
	int count = 0;
	unsigned long flags;
L
Linus Torvalds 已提交
5004 5005 5006 5007

	for (i = 0; i < conf->raid_disks; i++) {
		tmp = conf->disks + i;
		if (tmp->rdev
5008
		    && tmp->rdev->recovery_offset == MaxSector
5009
		    && !test_bit(Faulty, &tmp->rdev->flags)
5010
		    && !test_and_set_bit(In_sync, &tmp->rdev->flags)) {
5011
			count++;
5012
			sysfs_notify_dirent_safe(tmp->rdev->sysfs_state);
L
Linus Torvalds 已提交
5013 5014
		}
	}
5015 5016 5017
	spin_lock_irqsave(&conf->device_lock, flags);
	mddev->degraded -= count;
	spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
5018
	print_raid5_conf(conf);
5019
	return count;
L
Linus Torvalds 已提交
5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031
}

static int raid5_remove_disk(mddev_t *mddev, int number)
{
	raid5_conf_t *conf = mddev->private;
	int err = 0;
	mdk_rdev_t *rdev;
	struct disk_info *p = conf->disks + number;

	print_raid5_conf(conf);
	rdev = p->rdev;
	if (rdev) {
5032 5033 5034 5035
		if (number >= conf->raid_disks &&
		    conf->reshape_progress == MaxSector)
			clear_bit(In_sync, &rdev->flags);

5036
		if (test_bit(In_sync, &rdev->flags) ||
L
Linus Torvalds 已提交
5037 5038 5039 5040
		    atomic_read(&rdev->nr_pending)) {
			err = -EBUSY;
			goto abort;
		}
5041 5042 5043 5044
		/* Only remove non-faulty devices if recovery
		 * isn't possible.
		 */
		if (!test_bit(Faulty, &rdev->flags) &&
5045
		    mddev->recovery_disabled != conf->recovery_disabled &&
5046
		    !has_failed(conf) &&
5047
		    number < conf->raid_disks) {
5048 5049 5050
			err = -EBUSY;
			goto abort;
		}
L
Linus Torvalds 已提交
5051
		p->rdev = NULL;
5052
		synchronize_rcu();
L
Linus Torvalds 已提交
5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067
		if (atomic_read(&rdev->nr_pending)) {
			/* lost the race, try later */
			err = -EBUSY;
			p->rdev = rdev;
		}
	}
abort:

	print_raid5_conf(conf);
	return err;
}

static int raid5_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
{
	raid5_conf_t *conf = mddev->private;
5068
	int err = -EEXIST;
L
Linus Torvalds 已提交
5069 5070
	int disk;
	struct disk_info *p;
5071 5072
	int first = 0;
	int last = conf->raid_disks - 1;
L
Linus Torvalds 已提交
5073

5074 5075 5076
	if (mddev->recovery_disabled == conf->recovery_disabled)
		return -EBUSY;

5077
	if (has_failed(conf))
L
Linus Torvalds 已提交
5078
		/* no point adding a device */
5079
		return -EINVAL;
L
Linus Torvalds 已提交
5080

5081 5082
	if (rdev->raid_disk >= 0)
		first = last = rdev->raid_disk;
L
Linus Torvalds 已提交
5083 5084

	/*
5085 5086
	 * find the disk ... but prefer rdev->saved_raid_disk
	 * if possible.
L
Linus Torvalds 已提交
5087
	 */
5088
	if (rdev->saved_raid_disk >= 0 &&
5089
	    rdev->saved_raid_disk >= first &&
5090 5091 5092
	    conf->disks[rdev->saved_raid_disk].rdev == NULL)
		disk = rdev->saved_raid_disk;
	else
5093 5094
		disk = first;
	for ( ; disk <= last ; disk++)
L
Linus Torvalds 已提交
5095
		if ((p=conf->disks + disk)->rdev == NULL) {
5096
			clear_bit(In_sync, &rdev->flags);
L
Linus Torvalds 已提交
5097
			rdev->raid_disk = disk;
5098
			err = 0;
5099 5100
			if (rdev->saved_raid_disk != disk)
				conf->fullsync = 1;
5101
			rcu_assign_pointer(p->rdev, rdev);
L
Linus Torvalds 已提交
5102 5103 5104
			break;
		}
	print_raid5_conf(conf);
5105
	return err;
L
Linus Torvalds 已提交
5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116
}

static int raid5_resize(mddev_t *mddev, sector_t sectors)
{
	/* no resync is happening, and there is enough space
	 * on all devices, so we can resize.
	 * We need to make sure resync covers any new space.
	 * If the array is shrinking we should possibly wait until
	 * any io in the removed space completes, but it hardly seems
	 * worth it.
	 */
5117
	sectors &= ~((sector_t)mddev->chunk_sectors - 1);
5118 5119
	md_set_array_sectors(mddev, raid5_size(mddev, sectors,
					       mddev->raid_disks));
D
Dan Williams 已提交
5120 5121 5122
	if (mddev->array_sectors >
	    raid5_size(mddev, sectors, mddev->raid_disks))
		return -EINVAL;
5123
	set_capacity(mddev->gendisk, mddev->array_sectors);
5124
	revalidate_disk(mddev->gendisk);
5125 5126
	if (sectors > mddev->dev_sectors &&
	    mddev->recovery_cp > mddev->dev_sectors) {
A
Andre Noll 已提交
5127
		mddev->recovery_cp = mddev->dev_sectors;
L
Linus Torvalds 已提交
5128 5129
		set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
	}
A
Andre Noll 已提交
5130
	mddev->dev_sectors = sectors;
5131
	mddev->resync_max_sectors = sectors;
L
Linus Torvalds 已提交
5132 5133 5134
	return 0;
}

5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149
static int check_stripe_cache(mddev_t *mddev)
{
	/* Can only proceed if there are plenty of stripe_heads.
	 * We need a minimum of one full stripe,, and for sensible progress
	 * it is best to have about 4 times that.
	 * If we require 4 times, then the default 256 4K stripe_heads will
	 * allow for chunk sizes up to 256K, which is probably OK.
	 * If the chunk size is greater, user-space should request more
	 * stripe_heads first.
	 */
	raid5_conf_t *conf = mddev->private;
	if (((mddev->chunk_sectors << 9) / STRIPE_SIZE) * 4
	    > conf->max_nr_stripes ||
	    ((mddev->new_chunk_sectors << 9) / STRIPE_SIZE) * 4
	    > conf->max_nr_stripes) {
5150 5151
		printk(KERN_WARNING "md/raid:%s: reshape: not enough stripes.  Needed %lu\n",
		       mdname(mddev),
5152 5153 5154 5155 5156 5157 5158
		       ((max(mddev->chunk_sectors, mddev->new_chunk_sectors) << 9)
			/ STRIPE_SIZE)*4);
		return 0;
	}
	return 1;
}

5159
static int check_reshape(mddev_t *mddev)
5160
{
5161
	raid5_conf_t *conf = mddev->private;
5162

5163 5164
	if (mddev->delta_disks == 0 &&
	    mddev->new_layout == mddev->layout &&
5165
	    mddev->new_chunk_sectors == mddev->chunk_sectors)
5166
		return 0; /* nothing to do */
5167 5168 5169
	if (mddev->bitmap)
		/* Cannot grow a bitmap yet */
		return -EBUSY;
5170
	if (has_failed(conf))
5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183
		return -EINVAL;
	if (mddev->delta_disks < 0) {
		/* We might be able to shrink, but the devices must
		 * be made bigger first.
		 * For raid6, 4 is the minimum size.
		 * Otherwise 2 is the minimum
		 */
		int min = 2;
		if (mddev->level == 6)
			min = 4;
		if (mddev->raid_disks + mddev->delta_disks < min)
			return -EINVAL;
	}
5184

5185
	if (!check_stripe_cache(mddev))
5186 5187
		return -ENOSPC;

5188
	return resize_stripes(conf, conf->raid_disks + mddev->delta_disks);
5189 5190 5191 5192
}

static int raid5_start_reshape(mddev_t *mddev)
{
5193
	raid5_conf_t *conf = mddev->private;
5194 5195
	mdk_rdev_t *rdev;
	int spares = 0;
5196
	unsigned long flags;
5197

5198
	if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
5199 5200
		return -EBUSY;

5201 5202 5203
	if (!check_stripe_cache(mddev))
		return -ENOSPC;

5204
	list_for_each_entry(rdev, &mddev->disks, same_set)
5205 5206
		if (!test_bit(In_sync, &rdev->flags)
		    && !test_bit(Faulty, &rdev->flags))
5207
			spares++;
5208

5209
	if (spares - mddev->degraded < mddev->delta_disks - conf->max_degraded)
5210 5211 5212 5213 5214
		/* Not enough devices even to make a degraded array
		 * of that size
		 */
		return -EINVAL;

5215 5216 5217 5218 5219 5220
	/* Refuse to reduce size of the array.  Any reductions in
	 * array size must be through explicit setting of array_size
	 * attribute.
	 */
	if (raid5_size(mddev, 0, conf->raid_disks + mddev->delta_disks)
	    < mddev->array_sectors) {
5221
		printk(KERN_ERR "md/raid:%s: array size must be reduced "
5222 5223 5224 5225
		       "before number of disks\n", mdname(mddev));
		return -EINVAL;
	}

5226
	atomic_set(&conf->reshape_stripes, 0);
5227 5228
	spin_lock_irq(&conf->device_lock);
	conf->previous_raid_disks = conf->raid_disks;
5229
	conf->raid_disks += mddev->delta_disks;
5230 5231
	conf->prev_chunk_sectors = conf->chunk_sectors;
	conf->chunk_sectors = mddev->new_chunk_sectors;
5232 5233
	conf->prev_algo = conf->algorithm;
	conf->algorithm = mddev->new_layout;
5234 5235 5236 5237 5238
	if (mddev->delta_disks < 0)
		conf->reshape_progress = raid5_size(mddev, 0, 0);
	else
		conf->reshape_progress = 0;
	conf->reshape_safe = conf->reshape_progress;
5239
	conf->generation++;
5240 5241 5242 5243
	spin_unlock_irq(&conf->device_lock);

	/* Add some new drives, as many as will fit.
	 * We know there are enough to make the newly sized array work.
5244 5245 5246 5247
	 * Don't add devices if we are reducing the number of
	 * devices in the array.  This is because it is not possible
	 * to correctly record the "partially reconstructed" state of
	 * such devices during the reshape and confusion could result.
5248
	 */
5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260
	if (mddev->delta_disks >= 0) {
		int added_devices = 0;
		list_for_each_entry(rdev, &mddev->disks, same_set)
			if (rdev->raid_disk < 0 &&
			    !test_bit(Faulty, &rdev->flags)) {
				if (raid5_add_disk(mddev, rdev) == 0) {
					if (rdev->raid_disk
					    >= conf->previous_raid_disks) {
						set_bit(In_sync, &rdev->flags);
						added_devices++;
					} else
						rdev->recovery_offset = 0;
5261 5262

					if (sysfs_link_rdev(mddev, rdev))
5263
						/* Failure here is OK */;
5264
				}
5265 5266 5267 5268 5269 5270
			} else if (rdev->raid_disk >= conf->previous_raid_disks
				   && !test_bit(Faulty, &rdev->flags)) {
				/* This is a spare that was manually added */
				set_bit(In_sync, &rdev->flags);
				added_devices++;
			}
5271

5272 5273 5274 5275
		/* When a reshape changes the number of devices,
		 * ->degraded is measured against the larger of the
		 * pre and post number of devices.
		 */
5276
		spin_lock_irqsave(&conf->device_lock, flags);
5277
		mddev->degraded += (conf->raid_disks - conf->previous_raid_disks)
5278 5279 5280
			- added_devices;
		spin_unlock_irqrestore(&conf->device_lock, flags);
	}
5281
	mddev->raid_disks = conf->raid_disks;
5282
	mddev->reshape_position = conf->reshape_progress;
5283
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
5284

5285 5286 5287 5288 5289
	clear_bit(MD_RECOVERY_SYNC, &mddev->recovery);
	clear_bit(MD_RECOVERY_CHECK, &mddev->recovery);
	set_bit(MD_RECOVERY_RESHAPE, &mddev->recovery);
	set_bit(MD_RECOVERY_RUNNING, &mddev->recovery);
	mddev->sync_thread = md_register_thread(md_do_sync, mddev,
5290
						"reshape");
5291 5292 5293 5294
	if (!mddev->sync_thread) {
		mddev->recovery = 0;
		spin_lock_irq(&conf->device_lock);
		mddev->raid_disks = conf->raid_disks = conf->previous_raid_disks;
5295
		conf->reshape_progress = MaxSector;
5296 5297 5298
		spin_unlock_irq(&conf->device_lock);
		return -EAGAIN;
	}
5299
	conf->reshape_checkpoint = jiffies;
5300 5301 5302 5303 5304
	md_wakeup_thread(mddev->sync_thread);
	md_new_event(mddev);
	return 0;
}

5305 5306 5307
/* This is called from the reshape thread and should make any
 * changes needed in 'conf'
 */
5308 5309 5310
static void end_reshape(raid5_conf_t *conf)
{

5311 5312 5313
	if (!test_bit(MD_RECOVERY_INTR, &conf->mddev->recovery)) {

		spin_lock_irq(&conf->device_lock);
5314
		conf->previous_raid_disks = conf->raid_disks;
5315
		conf->reshape_progress = MaxSector;
5316
		spin_unlock_irq(&conf->device_lock);
5317
		wake_up(&conf->wait_for_overlap);
5318 5319 5320 5321

		/* read-ahead size must cover two whole stripes, which is
		 * 2 * (datadisks) * chunksize where 'n' is the number of raid devices
		 */
5322
		if (conf->mddev->queue) {
5323
			int data_disks = conf->raid_disks - conf->max_degraded;
5324
			int stripe = data_disks * ((conf->chunk_sectors << 9)
5325
						   / PAGE_SIZE);
5326 5327 5328
			if (conf->mddev->queue->backing_dev_info.ra_pages < 2 * stripe)
				conf->mddev->queue->backing_dev_info.ra_pages = 2 * stripe;
		}
5329 5330 5331
	}
}

5332 5333 5334
/* This is called from the raid5d thread with mddev_lock held.
 * It makes config changes to the device.
 */
5335 5336
static void raid5_finish_reshape(mddev_t *mddev)
{
5337
	raid5_conf_t *conf = mddev->private;
5338 5339 5340

	if (!test_bit(MD_RECOVERY_INTR, &mddev->recovery)) {

5341 5342 5343
		if (mddev->delta_disks > 0) {
			md_set_array_sectors(mddev, raid5_size(mddev, 0, 0));
			set_capacity(mddev->gendisk, mddev->array_sectors);
5344
			revalidate_disk(mddev->gendisk);
5345 5346 5347 5348 5349 5350 5351 5352 5353 5354
		} else {
			int d;
			mddev->degraded = conf->raid_disks;
			for (d = 0; d < conf->raid_disks ; d++)
				if (conf->disks[d].rdev &&
				    test_bit(In_sync,
					     &conf->disks[d].rdev->flags))
					mddev->degraded--;
			for (d = conf->raid_disks ;
			     d < conf->raid_disks - mddev->delta_disks;
5355 5356 5357
			     d++) {
				mdk_rdev_t *rdev = conf->disks[d].rdev;
				if (rdev && raid5_remove_disk(mddev, d) == 0) {
5358
					sysfs_unlink_rdev(mddev, rdev);
5359 5360 5361
					rdev->raid_disk = -1;
				}
			}
5362
		}
5363
		mddev->layout = conf->algorithm;
5364
		mddev->chunk_sectors = conf->chunk_sectors;
5365 5366
		mddev->reshape_position = MaxSector;
		mddev->delta_disks = 0;
5367 5368 5369
	}
}

5370 5371
static void raid5_quiesce(mddev_t *mddev, int state)
{
5372
	raid5_conf_t *conf = mddev->private;
5373 5374

	switch(state) {
5375 5376 5377 5378
	case 2: /* resume for a suspend */
		wake_up(&conf->wait_for_overlap);
		break;

5379 5380
	case 1: /* stop all writes */
		spin_lock_irq(&conf->device_lock);
5381 5382 5383 5384
		/* '2' tells resync/reshape to pause so that all
		 * active stripes can drain
		 */
		conf->quiesce = 2;
5385
		wait_event_lock_irq(conf->wait_for_stripe,
5386 5387
				    atomic_read(&conf->active_stripes) == 0 &&
				    atomic_read(&conf->active_aligned_reads) == 0,
5388
				    conf->device_lock, /* nothing */);
5389
		conf->quiesce = 1;
5390
		spin_unlock_irq(&conf->device_lock);
5391 5392
		/* allow reshape to continue */
		wake_up(&conf->wait_for_overlap);
5393 5394 5395 5396 5397 5398
		break;

	case 0: /* re-enable writes */
		spin_lock_irq(&conf->device_lock);
		conf->quiesce = 0;
		wake_up(&conf->wait_for_stripe);
5399
		wake_up(&conf->wait_for_overlap);
5400 5401 5402 5403
		spin_unlock_irq(&conf->device_lock);
		break;
	}
}
5404

5405

D
Dan Williams 已提交
5406
static void *raid45_takeover_raid0(mddev_t *mddev, int level)
5407
{
D
Dan Williams 已提交
5408
	struct raid0_private_data *raid0_priv = mddev->private;
5409
	sector_t sectors;
5410

D
Dan Williams 已提交
5411 5412
	/* for raid0 takeover only one zone is supported */
	if (raid0_priv->nr_strip_zones > 1) {
5413 5414
		printk(KERN_ERR "md/raid:%s: cannot takeover raid0 with more than one zone.\n",
		       mdname(mddev));
D
Dan Williams 已提交
5415 5416 5417
		return ERR_PTR(-EINVAL);
	}

5418 5419 5420
	sectors = raid0_priv->strip_zone[0].zone_end;
	sector_div(sectors, raid0_priv->strip_zone[0].nb_dev);
	mddev->dev_sectors = sectors;
D
Dan Williams 已提交
5421
	mddev->new_level = level;
5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432
	mddev->new_layout = ALGORITHM_PARITY_N;
	mddev->new_chunk_sectors = mddev->chunk_sectors;
	mddev->raid_disks += 1;
	mddev->delta_disks = 1;
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

	return setup_conf(mddev);
}


5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454
static void *raid5_takeover_raid1(mddev_t *mddev)
{
	int chunksect;

	if (mddev->raid_disks != 2 ||
	    mddev->degraded > 1)
		return ERR_PTR(-EINVAL);

	/* Should check if there are write-behind devices? */

	chunksect = 64*2; /* 64K by default */

	/* The array must be an exact multiple of chunksize */
	while (chunksect && (mddev->array_sectors & (chunksect-1)))
		chunksect >>= 1;

	if ((chunksect<<9) < STRIPE_SIZE)
		/* array size does not allow a suitable chunk size */
		return ERR_PTR(-EINVAL);

	mddev->new_level = 5;
	mddev->new_layout = ALGORITHM_LEFT_SYMMETRIC;
5455
	mddev->new_chunk_sectors = chunksect;
5456 5457 5458 5459

	return setup_conf(mddev);
}

5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492
static void *raid5_takeover_raid6(mddev_t *mddev)
{
	int new_layout;

	switch (mddev->layout) {
	case ALGORITHM_LEFT_ASYMMETRIC_6:
		new_layout = ALGORITHM_LEFT_ASYMMETRIC;
		break;
	case ALGORITHM_RIGHT_ASYMMETRIC_6:
		new_layout = ALGORITHM_RIGHT_ASYMMETRIC;
		break;
	case ALGORITHM_LEFT_SYMMETRIC_6:
		new_layout = ALGORITHM_LEFT_SYMMETRIC;
		break;
	case ALGORITHM_RIGHT_SYMMETRIC_6:
		new_layout = ALGORITHM_RIGHT_SYMMETRIC;
		break;
	case ALGORITHM_PARITY_0_6:
		new_layout = ALGORITHM_PARITY_0;
		break;
	case ALGORITHM_PARITY_N:
		new_layout = ALGORITHM_PARITY_N;
		break;
	default:
		return ERR_PTR(-EINVAL);
	}
	mddev->new_level = 5;
	mddev->new_layout = new_layout;
	mddev->delta_disks = -1;
	mddev->raid_disks -= 1;
	return setup_conf(mddev);
}

5493

5494
static int raid5_check_reshape(mddev_t *mddev)
5495
{
5496 5497 5498 5499
	/* For a 2-drive array, the layout and chunk size can be changed
	 * immediately as not restriping is needed.
	 * For larger arrays we record the new value - after validation
	 * to be used by a reshape pass.
5500
	 */
5501
	raid5_conf_t *conf = mddev->private;
5502
	int new_chunk = mddev->new_chunk_sectors;
5503

5504
	if (mddev->new_layout >= 0 && !algorithm_valid_raid5(mddev->new_layout))
5505 5506
		return -EINVAL;
	if (new_chunk > 0) {
5507
		if (!is_power_of_2(new_chunk))
5508
			return -EINVAL;
5509
		if (new_chunk < (PAGE_SIZE>>9))
5510
			return -EINVAL;
5511
		if (mddev->array_sectors & (new_chunk-1))
5512 5513 5514 5515 5516 5517
			/* not factor of array size */
			return -EINVAL;
	}

	/* They look valid */

5518
	if (mddev->raid_disks == 2) {
5519 5520 5521 5522
		/* can make the change immediately */
		if (mddev->new_layout >= 0) {
			conf->algorithm = mddev->new_layout;
			mddev->layout = mddev->new_layout;
5523 5524
		}
		if (new_chunk > 0) {
5525 5526
			conf->chunk_sectors = new_chunk ;
			mddev->chunk_sectors = new_chunk;
5527 5528 5529
		}
		set_bit(MD_CHANGE_DEVS, &mddev->flags);
		md_wakeup_thread(mddev->thread);
5530
	}
5531
	return check_reshape(mddev);
5532 5533
}

5534
static int raid6_check_reshape(mddev_t *mddev)
5535
{
5536
	int new_chunk = mddev->new_chunk_sectors;
5537

5538
	if (mddev->new_layout >= 0 && !algorithm_valid_raid6(mddev->new_layout))
5539
		return -EINVAL;
5540
	if (new_chunk > 0) {
5541
		if (!is_power_of_2(new_chunk))
5542
			return -EINVAL;
5543
		if (new_chunk < (PAGE_SIZE >> 9))
5544
			return -EINVAL;
5545
		if (mddev->array_sectors & (new_chunk-1))
5546 5547
			/* not factor of array size */
			return -EINVAL;
5548
	}
5549 5550

	/* They look valid */
5551
	return check_reshape(mddev);
5552 5553
}

5554 5555 5556
static void *raid5_takeover(mddev_t *mddev)
{
	/* raid5 can take over:
D
Dan Williams 已提交
5557
	 *  raid0 - if there is only one strip zone - make it a raid4 layout
5558 5559 5560 5561
	 *  raid1 - if there are two drives.  We need to know the chunk size
	 *  raid4 - trivial - just use a raid4 layout.
	 *  raid6 - Providing it is a *_6 layout
	 */
D
Dan Williams 已提交
5562 5563
	if (mddev->level == 0)
		return raid45_takeover_raid0(mddev, 5);
5564 5565
	if (mddev->level == 1)
		return raid5_takeover_raid1(mddev);
5566 5567 5568 5569 5570
	if (mddev->level == 4) {
		mddev->new_layout = ALGORITHM_PARITY_N;
		mddev->new_level = 5;
		return setup_conf(mddev);
	}
5571 5572
	if (mddev->level == 6)
		return raid5_takeover_raid6(mddev);
5573 5574 5575 5576

	return ERR_PTR(-EINVAL);
}

5577 5578
static void *raid4_takeover(mddev_t *mddev)
{
D
Dan Williams 已提交
5579 5580 5581
	/* raid4 can take over:
	 *  raid0 - if there is only one strip zone
	 *  raid5 - if layout is right
5582
	 */
D
Dan Williams 已提交
5583 5584
	if (mddev->level == 0)
		return raid45_takeover_raid0(mddev, 4);
5585 5586 5587 5588 5589 5590 5591 5592
	if (mddev->level == 5 &&
	    mddev->layout == ALGORITHM_PARITY_N) {
		mddev->new_layout = 0;
		mddev->new_level = 4;
		return setup_conf(mddev);
	}
	return ERR_PTR(-EINVAL);
}
5593

5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642
static struct mdk_personality raid5_personality;

static void *raid6_takeover(mddev_t *mddev)
{
	/* Currently can only take over a raid5.  We map the
	 * personality to an equivalent raid6 personality
	 * with the Q block at the end.
	 */
	int new_layout;

	if (mddev->pers != &raid5_personality)
		return ERR_PTR(-EINVAL);
	if (mddev->degraded > 1)
		return ERR_PTR(-EINVAL);
	if (mddev->raid_disks > 253)
		return ERR_PTR(-EINVAL);
	if (mddev->raid_disks < 3)
		return ERR_PTR(-EINVAL);

	switch (mddev->layout) {
	case ALGORITHM_LEFT_ASYMMETRIC:
		new_layout = ALGORITHM_LEFT_ASYMMETRIC_6;
		break;
	case ALGORITHM_RIGHT_ASYMMETRIC:
		new_layout = ALGORITHM_RIGHT_ASYMMETRIC_6;
		break;
	case ALGORITHM_LEFT_SYMMETRIC:
		new_layout = ALGORITHM_LEFT_SYMMETRIC_6;
		break;
	case ALGORITHM_RIGHT_SYMMETRIC:
		new_layout = ALGORITHM_RIGHT_SYMMETRIC_6;
		break;
	case ALGORITHM_PARITY_0:
		new_layout = ALGORITHM_PARITY_0_6;
		break;
	case ALGORITHM_PARITY_N:
		new_layout = ALGORITHM_PARITY_N;
		break;
	default:
		return ERR_PTR(-EINVAL);
	}
	mddev->new_level = 6;
	mddev->new_layout = new_layout;
	mddev->delta_disks = 1;
	mddev->raid_disks += 1;
	return setup_conf(mddev);
}


5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657
static struct mdk_personality raid6_personality =
{
	.name		= "raid6",
	.level		= 6,
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid5_add_disk,
	.hot_remove_disk= raid5_remove_disk,
	.spare_active	= raid5_spare_active,
	.sync_request	= sync_request,
	.resize		= raid5_resize,
5658
	.size		= raid5_size,
5659
	.check_reshape	= raid6_check_reshape,
5660
	.start_reshape  = raid5_start_reshape,
5661
	.finish_reshape = raid5_finish_reshape,
5662
	.quiesce	= raid5_quiesce,
5663
	.takeover	= raid6_takeover,
5664
};
5665
static struct mdk_personality raid5_personality =
L
Linus Torvalds 已提交
5666 5667
{
	.name		= "raid5",
5668
	.level		= 5,
L
Linus Torvalds 已提交
5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid5_add_disk,
	.hot_remove_disk= raid5_remove_disk,
	.spare_active	= raid5_spare_active,
	.sync_request	= sync_request,
	.resize		= raid5_resize,
5680
	.size		= raid5_size,
5681 5682
	.check_reshape	= raid5_check_reshape,
	.start_reshape  = raid5_start_reshape,
5683
	.finish_reshape = raid5_finish_reshape,
5684
	.quiesce	= raid5_quiesce,
5685
	.takeover	= raid5_takeover,
L
Linus Torvalds 已提交
5686 5687
};

5688
static struct mdk_personality raid4_personality =
L
Linus Torvalds 已提交
5689
{
5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702
	.name		= "raid4",
	.level		= 4,
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid5_add_disk,
	.hot_remove_disk= raid5_remove_disk,
	.spare_active	= raid5_spare_active,
	.sync_request	= sync_request,
	.resize		= raid5_resize,
5703
	.size		= raid5_size,
5704 5705
	.check_reshape	= raid5_check_reshape,
	.start_reshape  = raid5_start_reshape,
5706
	.finish_reshape = raid5_finish_reshape,
5707
	.quiesce	= raid5_quiesce,
5708
	.takeover	= raid4_takeover,
5709 5710 5711 5712
};

static int __init raid5_init(void)
{
5713
	register_md_personality(&raid6_personality);
5714 5715 5716
	register_md_personality(&raid5_personality);
	register_md_personality(&raid4_personality);
	return 0;
L
Linus Torvalds 已提交
5717 5718
}

5719
static void raid5_exit(void)
L
Linus Torvalds 已提交
5720
{
5721
	unregister_md_personality(&raid6_personality);
5722 5723
	unregister_md_personality(&raid5_personality);
	unregister_md_personality(&raid4_personality);
L
Linus Torvalds 已提交
5724 5725 5726 5727 5728
}

module_init(raid5_init);
module_exit(raid5_exit);
MODULE_LICENSE("GPL");
5729
MODULE_DESCRIPTION("RAID4/5/6 (striping with parity) personality for MD");
L
Linus Torvalds 已提交
5730
MODULE_ALIAS("md-personality-4"); /* RAID5 */
5731 5732
MODULE_ALIAS("md-raid5");
MODULE_ALIAS("md-raid4");
5733 5734
MODULE_ALIAS("md-level-5");
MODULE_ALIAS("md-level-4");
5735 5736 5737 5738 5739 5740 5741
MODULE_ALIAS("md-personality-8"); /* RAID6 */
MODULE_ALIAS("md-raid6");
MODULE_ALIAS("md-level-6");

/* This used to be two separate modules, they were: */
MODULE_ALIAS("raid5");
MODULE_ALIAS("raid6");